Classification-based melody transcription
- 格式:pdf
- 大小:918.43 KB
- 文档页数:20
基因英语词汇翻译Aactivation domain 活化结构域adapters 连接物adenine 腺嘌呤adenosine 腺ADP (adenosine diphosphate) 腺二磷酸affinity column 亲和柱AFLP (amplified fragment length polymorphisms) 增值性断片长度多态现象agrobacterium 农杆菌属alanine 丙氨酸allele 等位基因amber mutation 琥珀型突变AMP (adenosine monophosphate) 腺一磷酸ampicillin 氨?青霉素anchor primer 锚状引物annealing 退火annealing temperature 退火温度anticodon 反密码子AP-PCR (arbitrarily primed PCR) 任意引物聚合?链反应arbitrary primer 任意引物ATP (adenosine triphosphate) 腺三磷酸autosome 常染色体腺苷脱氨酶缺乏症 adenosine deaminasedeficiency (ADA) 腺病毒 adenovirusAlagille综合征 Alagille syndrome等位基因 allele氨基酸 amino acids动物模型 animal model抗体 antibody凋亡 apoptosis路-巴综合征ataxia-telangiectasia常染色体显性autosomal dominant常染色体 autosomeBbaculovirus 杆状病毒base pair ..基对base sequence ..基顺序beta-galactosidase ..-半乳糖? beta-glucuronidase ..-葡糖醛酸糖? bioluminescence 生物发光bioremediation 生物降解biotechnology 生物技术blotting 印迹法blue-white selection 蓝白斑筛选细菌人工染色体 bacterial artificial chromosome (BAC)碱基对 base pair先天缺陷birth defect骨髓移植bone marrow transplantation blunt end 平(整末)端Ccatalyst 催化剂cDNA library 反向转录DNA库centromere 着丝体centrosome 中心体chemiluminescence 化学发光chiasma 交叉chromomere 染色粒chromoplast 有色体chromosomal aberration 染色体畸变chromosomal duplication 染色体复制chromosomal fibre 染色体牵丝chromosome 染色体chromosome complement 染色体组chromosome map 染色体图chromosome mutation 染色体突变clone 克隆cloning 无性繁殖系化codon 密码子codon degeneracy 密码简并codon usage 密码子选择cohesive end 黏性末端complementary DNA (cDNA) 反向转录DNA complementary gene 互补基因consensus sequence 共有序列construct 组成cosmids 黏性质粒crossing over 互换cyclic AMP (cAMP) 环腺酸cytosine 胞嘧啶癌 cancer后选基因 candidate gene癌 carcinomacDNA文库 cDNA library 细胞cell染色体 chromosome克隆 cloning密码 codon天生的 congenital重叠群 contig囊性纤维化 cystic fibrosis 细胞遗传图 cytogenetic mapDdark band 暗带deamination 脱氨基作用decarboxylation 脱羧基作用degenerate code 简并密码degenerate PCR 退化性聚合?链反应dehydrogenase 脱氢?denaturation 变性deoxyribonucleoside diphospahte 脱氧核糖核一磷酸deoxyribonucleoside monophospahte 脱氧核糖核二磷酸deoxyribonucleoside triphospahte 脱氧核糖核三磷酸deoxyribose 去(脱)氧核糖dicarboxylic acid 二羧酸digoxigenin 洋地黄毒diploid 二倍体DNA (deoxyribonucleic acid) 去(脱)氧核糖核酸DNA binding domain DNA结合性结构域DNA fingerprinting DNA指纹图谱DNA helicase DNA解螺旋?DNA kinase DNA激?DNA ligase DNA连接?DNA polymer DNA聚合物DNA polymerase DNA聚合?double helix 双螺旋double-strand 双链缺失 deletion脱氧核糖核酸 deoxyribonucleic acid (DNA) 糖尿病 diabetes mellitus二倍体 diploidDNA复制 DNA replicationDNA测序 DNA sequencing显性的 dominant双螺旋 double helix复制 duplicationEelectroporation 电穿孔endonuclease 内切核酸? enhancer 增强子enterokinase 肠激? episome 游离基因ethidium bromide 溴乙锭eukaryotic 真核生物的euploid 整倍体exonuclease 外切核酸?expressed-sequence tags 表达的序列标记片段extron 外含子电泳electrophoresis 酶enzyme外显子exonFF factor F因子FAD (flavine adenine dinucleotide) 黄素腺嘌呤二(双)核酸feedback control 反馈控制feedback inhibition 反馈抑制feedback mechanism 反馈机制first filial (F1) generation 第一子代FISH (fluoresence in situ hybridization) 荧光原位杂交forward mutation 正向突变F-pilus F纤毛functional complementation 功能性互补作用fusion protein 融合蛋白家族性地中海热familial Mediterraneanfever 荧光原位杂交fluorescence in situhybridization (FISH) 脆性X染色体综合征Fragile X syndromeGgel electrophoresis 凝胶电泳gene 基因gene cloning 基因克隆gene conversion 基因转变gene duplication 基因复制gene flow 基因流动gene gun 基因枪gene interaction 基因相互作用gene locus 基因位点gene mutation 基因突变gene regulation 基因调节gene segregation 基因分离gene therapy 基因治疗geneome 基因组/ 染色体组genetic map 基因图genetic modified foods (GM foods) 基因食物genetics 遗传学genetypic ratio 基因型比/ 基因型比值genome 基因组/ 染色体组genomic library 基因组文库genotype 基因型giant chromosome 巨染色体globulin 球蛋白glucose-6-phosphate dehydrogenase 6-磷酸葡萄糖脱氢?GP (glycerate phosphate) 磷酸甘油酸脂GTP (guanine triphosphate) 鸟三磷酸guanine 鸟嘌呤基因扩增gene amplification基因表达gene expression基因图谱gene mapping基因库gene pool基因治疗gene therapy基因转移gene transfer遗传密码genetic code (A TGC)遗传咨询genetic counseling遗传图genetic map遗传标记genetic marker遗传病筛查genetic screening基因组genome基因型genotype种系germ lineHhaploid 单倍体haploid generation 单倍世代heredity 遗传heterochromatin 异染色质Hfr strain 高频重组菌株holoenzyme 全?homologous 同源的housekeeping gene 家务基因hybridization 杂交单倍体haploid造血干细胞hematopoietic stem cell 血友病hemophilia 杂合子heterozygous高度保守序列highly conserved sequence Hirschsprung病Hirschsprung's disease纯合子homozygous人工染色体human artificial chromosome (HAC)人类基因组计划Human Genome Project human immunodeficiency virus (HIV)/ 人类免疫缺陷病毒acquired immunodeficiency syndrome (AIDS) 获得性免疫缺陷综合征huntington舞蹈病Huntington's diseaseIimmunoglobulin 免疫球蛋白in vitro 在体外/ 在试管内in vivio 在体内independent assortment 独立分配induced mutation 诱发性突变induction 诱导initiation codon 起始密码子inosine 次黄insert 插入片段insertional inactivation 插入失活interference 干扰intergenic 基因间的interphase 间期intragenic 基因内的intron 内含子inversion 倒位isocaudarner 同尾酸isoschizomer 同切点?Kkanamycin 卡那毒素klenow fragment 克列诺夫片段Llac operon 乳糖操纵子ligase 连接? ligation 连接作用light band 明带linker 连接体liposome 脂质体locus 位点Mmap distance 图距离map unit 图距单位mature transcript 成熟转录物metaphase 中期methylase 甲基化? methylation 甲基化作用microarray 微列microinjection 微注射missense mutation 错差突变molecular genetics 分子遗传学monoploid 单倍体monosome 单染色体messenger RNA (mRNA) 信使RNA multiple alleles 复(多)等位基因mutagen 诱变剂mutagenesis 诱变mutant 突变体mutant gene 突变基因mutant strain 突变株mutation 突变mutation rate 突变率muton 突变子畸形malformation描图mapping标记marker黑色素瘤melanoma孟德尔Mendel, Johann (Gregor)孟德尔遗传Mendelian inheritance信使RNA messenger RNA (mRNA)[分裂]中期metaphase微阵技术microarray technology线立体DNA mitochondrial DNA单体性monosomy小鼠模型mouse model多发性内分泌瘤病multiple endocrine neoplasia, type 1 (MEN1)NNAD (nicotinamide adenine dinucleotide) 烟醯胺腺嘌呤二核酸NADP (nicotinamide adenine dinucleotide phosphate) 烟醯胺腺嘌呤二核酸磷酸nicking activity 切割活性nonsense codon 无意义密码子nonsense mutation 无意义突变Northern blot Northern印迹法nuclear DNA 核DNAnuclear gene 核基因nuclease 核酸?nucleic acid 核酸nucleoside 核nucleoside triphosphate 核三磷酸nucleotidase 核酸?nucleotide 核酸nucleotide sequence 核酸序列神经纤维瘤病neurofibromatosis尼曼-皮克病Niemann-Pick disease, type C (NPC)RNA印记Northern blot核苷酸nucleotide神经核nucleusOoligonucleotide 寡核酸one gene one polypeptide hypothesis 一个基因学说operon 操纵子oxidative decarboxylation 氧化脱羧作用oxidative phosphorylation 氧化磷酸化作用寡核苷酸oligo癌基因oncogenePpeptide ? peptide bond ?键phagemids 噬菌粒phosphorylation 磷酸化作用physical map 物理图谱plasmid 质粒point mutation 点突变poly(A) tail poly(A)尾polymerase 聚合?polyploid 多倍体positional cloning 位置性无性繁殖系化primary transcript 初级转录物primer 引物probe 探针prokaryotic 原核的promoter 启动子protease 蛋白?purine 嘌呤pyrimidine 嘧啶Parkinson病Parkinson's disease血系/谱系pedigree表型phenotype物理图谱physical map多指畸形/多趾畸形polydactyly聚合酶链反应polymerase chain reaction (PCR)多态性polymorphism定位克隆positional cloning原发性免疫缺陷primary immunodeficiency 原核pronucleus前列腺癌prostate cancerRrandom segregation 随机分离RAPD (rapid amplified polymorphic DNA) 快速扩增多态DNAreading frame 阅读码框recessive gene 隐性基因recombinant 重组体recombinant DNA technology 重组DNA技术recombination 重组regulator (gene) 调控基因replica 复制物/ 印模replica plating 复制平皿(板)培养法replication 复制replication origin 复制起点reporter gene 报道基因repression 阻遏repressor 阻遏物repressor gene 阻遏基因resistance strain 抗药性菌株restriction 限制作用restriction enzyme 限制性内切? restriction mapping 限制性内切?图谱retrovirus 反转录病毒reverse transcription 反转录作用RFLP (restricted fragment length polymorphisms) 限制性断片长度多态现象ribonucleotide 核糖核酸ribose 核糖ribosomal RNA (rRNA) 核糖体RNA ribosome 核糖体RNA (ribonucleic acid) 核糖核酸RNA polymerase I RNA聚合?IRNA polymerase II RNA聚合?IIRNA polymerase III RNA聚合?IIIR-plasmid R质粒/ 抗药性质粒隐性recessive逆转录病毒retrovirus核糖核酸ribonucleic acid (RNA)核糖体ribosomeSsecond filial (F2) generation 第二子代self-ligation 自我连接作用shuttle vectors 穿梭载体sigma factor ..因子single nucleotide polymorphism 单核酸多态性single-stranded DNA 单链DNAsister chromatid 姊妹染色单体sister chromosome 姊妹染色体site-directed mutagenesis 定点诱变somatic cell 体细胞Southern blot Southern印迹法splice 拼接star activity 星号活性stationary phase 静止生长期sticky end 黏性末端stop codon 终止密码子structural gene 结构基因supernatant 上层清液supressor 抑制基因序列标记位点sequence-tagged site (STS) 联合免疫缺陷severe combined immunodeficiency (SCID)性染色体sex chromosome伴性的sex-linked体细胞somatic cellsDNA印记Southern blot光谱核型spectral karyotype (SKY)替代substitution自杀基因suicide gene综合征syndromeTtelophase 末期template 模板terminator 终止子tetracycline 四环素thymine 胸腺嘧啶tissue culture 组织培养transcription 转录作用transfer RNA (tRNA) 转移RNA transformation 转化作用transgene 转基因translation 翻译/ 平移transmembrane 跨膜triplet 三联体triplet code 三联体密码triploid 三倍体技术转让technology transfer转基因的transgenic易位translocation三体型trisomy肿瘤抑制基因tumor suppressor geneVvector 载体WWestern blot Western印迹法Wolfram综合征Wolfram syndromeY 酵母人工染色体yeast artificial chromosome (YAC)。
Alphabet n. 字母Abandon 抛弃BandBond 纽带联系BendBend overBoundBe bound to HusbandHouseHousewife HouseworkBan-bannedBunchA bunch of AbundantApart 离开PartAbroad-go abroadgo aboardBroadAroundRound Roundabout surround围绕SurroundingAcrossCrossCrossing CrossroadsRoadRing-road SideroadSidewalkRoutineIn-insideEx-exitSuper-supermanSub-subwayEnjoy-joy Compass指南针-pass Research-search AbleUnableDisabled 残疾Ability能力Disability EnjoyableReasonableReasonChangeableChangeExchangeExchange student交换生Admirable 值得赞赏的、很好的Admire 钦佩、羡慕MirrorMiracleComfortableUncomfortableComfort zone 舒适区It takes time andeffort to do a goodjob.想做好一件事,需要花时间和努力。
Fore sb. To do sth 强迫某人做某事WorkforceFierce 猛烈A fierce snow storm一场猛烈的暴风雪Unbelievable 难以置信的Believe 相信BeliefCreditThose wereincredible days 那些日子真的太棒了Impossible 不可能的Possible 可能的PossilbilityPossiblyPotential market 潜在市场Abnormal = that isnot normalNormalAbove 在上面Below 在下面These new chances belong to theyounger generation.新的机会,属于年轻一代Belong to 属于Aboard上(船、车、飞机等)Board 木板Border 边境,国界Bridge 桥Skateboard冰鞋Skate溜冰Ski 滑雪SkillSkilledSkillfulSkillfullyI am absent from the meeting,我缺席了会议Absent 缺席Absence 缺席Present 出现、呈现Presentation 演示PreparePreparationPresident 总统RepresentRepresentativeAbrupt 突然的、意外的Interrupt 打断Our life is interrupted.我们的生活被打扰了Internet 互联网Net 网Corrupt 贪污的Erupt 火山爆发Arid 干旱A friendly and relaxedAtmosphere 大气层、氛围balmy温和的= mild = temperate= moderateBarometer气压计= indicator指示器Blast一阵风= gust = blowBlizzard 大风雪Breeze 微风Chill 变冷chilly寒冷的ClimateWeatherAtmosphereCondense浓缩Convection zone 对流层Crystal 水晶Current 气流Humid= Damp 潮湿= dank = clammy = humidity = moisture Dew 露水Downpour倾盆大雨Drizzle 下细雨= rain = sprinkle droplet小滴Drought 干旱= aridity= dry period = prolonged lack of rain Frigid 严寒的Frost 霜Funnel 漏斗云Gale大风Greenhouse effect 温室效应Hail 冰雹Hurricane 飓风Meteorology 气象学Moisture 潮湿Oxygen氧气Ozone layer 臭氧层Precipitate 加速Precipitation 降水Saturate 饱和Fog 雾Serene 平静的Smog 烟雾Temperature 温度Tempest 暴风雨Tepid = lukewarm = warm 微温的Tornado龙卷风Troposphere 对流层Typhoon 台风Vapor 水蒸气Whirlwind 旋风Christianity 基督教Catholicism 天主教Protestantism 新教Reformation 改革Lutheranism 路德教Calvinism 加尔文教派Methodism 卫理公会Puritanism 清教Judaism 犹太教Islamism 伊斯兰教Buddhism 佛教Daoism 道教Paganism 拜物主义Atheism 无神论Mammal哺乳动物Buffalo水牛Calf 小牛Zebra斑马Antelope 羚羊Gazelle瞪羚Reindeer 驯鹿Dromedary 单峰骆驼Rhinoceros 犀牛Leopard 豹子Beaver 狸Chimpanzee 黑猩猩Gorilla大猩猩Hedgehog 刺猬Walrus 海象Eagle 鹰Falcon猎鹰Vulture 贪婪的人Turkey 火鸡Peacock孔雀Ostrich 鸵鸟Canary 金丝雀Reptile 爬行动物、卑鄙的人Batrachia蛙类Python 巨蟒Rettlesnake 响尾蛇Lizard 蜥蜴Chameleon变色龙Crocodile 鳄鱼Turtle 海龟Salmon 鲑鱼Sardine 沙丁鱼Cicada蟋蟀Centipede 蜈蚣Butterfly蝴蝶Scorpion 蝎子Mollusk 软体动物Cuttlefish 墨鱼Octopus 章鱼Lobster 龙虾Prawn 大虾Worm虫Earthworm 蚯蚓Baboon 狒狒Moth 蛀虫Caterpillar 毛虫Dinosaur 恐龙Larva 幼虫Family 科Class纲Order 目Suborder 亚目Genus 种类Antenna 触须Tentacle 触角Spleen 脾脏Hide 兽皮Spine 脊骨Toe脚趾Bill 鸟嘴Beak鸟嘴Fuzzy有绒毛的Scale鳞片Nervous 神经的Grease 脂肪Jellyfish 水母Starfish 海星Porpoise 海豚Shrimp 小虾Sponge 海绵Plankton 浮游生物Oyster 生蚝Clam 蛤蜊Coral 珊瑚虫Crab螃蟹Reagent反应物Element 元素Compound 化合物Molecule 分子Electron 电子Isotope 同位素Polymer 聚合体Alloy合金Metal 金属Metalloid 非金属Derivative 衍生物Alkali碱性的Hydrate 水合物Action 作用Adhesive 粘合剂Alchemy 炼金术Biochemistry 生物化学Bleach 漂白Blast爆破Calcium钙Carbon 碳Catalysis 催化作用Caustic 腐蚀性的= abrasive = corrosiveCombination 组合Corrode腐蚀= erode = wat awayExplode 爆炸Explosive 爆炸的、炸药Gasoline汽油Ignite = inflame = kindle 燃Impurity 杂志Iodine 碘酒Methane 沼气Nickle 镍Nitrogen 氮Particle 微小的颗粒Polymerization n.聚合Scorch 使褪色Silicon 硅Sodium 纳Solubility 溶度Solvent 溶解Sulfur 硫磺Synthetic 综合的Tint 上色Zinc 锌Solution 解决办法Rotten 腐烂的Complusory 必须的、强迫的Optional 任选的Elective 可选修的Obligatory 必修的Socratic 苏格拉底式的Subject 科目、学科Discipline 纪律、学科Interdisciplinary 各学科间的Instruct 教导、命令Enlighten 启迪、教化Curriculum 课程Mathematics 数学Science 科学Arts文科Literacy 有文化Illiteracy 文盲Primary 初步的、初级的Secondary 二级的、中级的Tertiary 高等的、第三级的Matriculate 被录取入学Enroll 登记、招收Admission 准许进入Inculcate 谆谆教诲Rote learning 机械学习Credit 学分Semester 学期Specialization 专门化、专业Major 主修课Minor 副修科目Mobility 活动性Internationalization 国际化Confer 授予Award 授予、奖学金Specialty 专业Academia学术界Symposium n.讨论会、座谈会Autonomy n.自治Heuristic 启发式的Elicitation 引出、诱出、启发Universal 普遍的、全体的Elementary 初步的、基本的V ocational 职业的Faculty 才能、教员Chancellor 大学校长Extracurricular 课外的、业余的Alumna 女校友Alumnus男校友Transcript 成绩单Dissertation 论文、专题Thesis 论题、论文Scholarship 奖学金Tuition 学费Accommodation 住宿Allowance 津贴补助Didactic 说教的= instructiveProductionSavingInvestmentExpenditureCapitalCurrencyDenomination n.(种类、数值、大小)单位Demand n.需求、需要Supply n.补给、供应品Purchase 购买= buyInflation 通货膨胀Monetary 货币的、金钱的Microeconomics微观经济学Macroeconomics宏观经济学Consume = spend 消费Distribution 分配、分发Durability 持久力Nominal 名义上的Taxation 课税Externality n. 外部因素Asymmetric 不均匀的、不对称的Fluctuate 变动、波动Predictability 可预见性Trademark 商标Copyright 版权著作Brand 商标Appreciation 增殖Depreciation 贬值Depression 萧条不景气Discount 折扣Welfare 福利、安宁Efficiency 效率Unemployment 失业Entrepreneur 企业家Factor 因素Reserve 储存Fiscal财政的、会计的Industrialization 工业化Deflation 通货紧缩Index 指标Securities 证券Insurance 保险Futures 期货Inventory 存货、盘点Financial 财政的Budget n. 预算Equilibrium 平衡、平静Hyperinflation 恶性通货膨胀Aggregate 合计、总计Incentive 刺激Profitability 收益性、利率Allocation n. 分配Rigidity 分配、安置Share 共享、份额Stock n.股票Payment 付款Stagger摇晃Distortion扭曲、变形Function 功能、作用Tariff关税Quota 配额Textile 纺织品Commodity 日用品、商品Poverty 贫困、贫乏Scarcity 缺乏不足Dearth 缺乏Surplus 剩余Privatize 私有Affluence 富裕、富足Prosperity 繁荣Deprivation 剥夺Paucity 缺乏Plethora 过剩、过多Abundace 丰富、充裕Compensate 补偿、偿还Merchandise 商品Enterprise 企业、事业Commerce 商业Discount 折扣Account 计算、账目Collateral 担保的Reimburse 偿还Refund 偿还、退款Transaction n.交易、事务Patronage 赞助Sponsorship 赞助Benefaction n.恩惠、善行Commission n.佣金Ransom 赎金Interest 利息、利益Bankruptcy 破产Acquisition 收购、获得Merge 合并Confidential 机密的Classified 机密的Military 军事的、军用的Martial 战争的、军事的Navy 海军Armada 舰队Flotilla 小舰队Fleet 舰队Campaign 战役、作战Diplomacy 外交Strategy 策略Cannon 大炮Radar 雷达Morale 民心Besiege 围Blockade 阻塞Expedition n.远征Mission 使命、任务Repulse 击退Rebuff挫败Revolt 反抗、起义Rebellion 谋反、反抗Revolution 革命Insurrection 起义Mutiny 兵变、反抗Riot 暴乱envelop包围Encircle 环绕Invade 入侵Wreck 破坏Demolish = dismantle = shatter毁坏、破坏Collision 冲突Debris 碎片Encroach 蚕食、侵占Exterminate 消灭Trespass 侵入Enlist 征召Disarming 消除敌意的Assault 攻击、袭击Offense 进攻Defense 防御Onslaught 猛烈攻击Armament 兵力、军力Disarm 消除(敌意)Neutralize 使中立化Captivate 抓住、捕获Fortress 堡垒Weapon 武器Destruction 破坏Warfare 战争、作战、冲突Rivalry 竞争Masterpiece 杰作、著作Gallery 画廊Exhibition 展览会Collection 收藏Inspiration 灵感Purism 纯粹主义Byzantine 拜占庭画家Surrealism 超现实主义Classicism 古典主义Baroque 巴洛克式Rococo 过分精巧的Impressionism 印象流派Literature 文学Folklore 民间传说Essay 散文Criticism 批评Anthology 诗选V olume 卷、册Drama 戏剧Comedy 喜剧Tragedy 悲剧Playwright剧作家Episode 插曲Biography 传记Improvisation 即席创作Eloquence 雄辩、口才Pigment 色素Portrait 肖像Caricature 讽刺画Easel 画架Bronze 铜像Sculpture 雕塑品Architecture 建筑、建筑学Contaminate= defile=pollute 污染Taint 腐坏Purify纯净Pollutant 污染杂质Ecosystem 生态系统Noxious 有害的、有毒的Toxic 有毒的、中毒的Lethal 致命的= fatalVenomous 有毒的Deteriorate 恶化Pernicious 有害的Decibel 分贝Hazardous 危险的Effluent 排污Emission 发射Inorganic 无机的Litter垃圾Nuisance =pest令人讨厌的东西Radioactive 放射性的Sewage 下水道ConservationRenewableAtmosphereForestationDesertificationPreservationReservoirOasisErrorAccuracyAngleFormulaFunctionAdditionDivideMultipleSubtract=deductAdjacent=neighboringAlgebraArithmeticEquationAltitudegeometry几何学Circumference周围Slope 斜坡Summation 总和Symmetry 对称Approximate=roughlySequence= procession =progressionProgressionAscendingAssumption=suppositionConclusionProbabilityCalculationClassificationCombinationLogarithmComplementarySquareCongruentbeautyConstantCoordinateCylinder=columnDenominatorDeviationDimensionEvaluate=estimate=assessHemisphereHorizontalHyperbolaIntersect=cross=meetIn+vari(变)+ance=n.不变性Increment 增加Maximize最大值Negative Numerator Dividend Pentagon Perimeter Permutation Circumference Permutation Quadrant QuotientFractionDecimal DiameterEllipseRadiusVertical Deduction StatisticsEven numbers 偶数Odd奇数Enumerate 计数Calculate =count Calculus Percentage Proportion ExponentIndexDerivativepower幂Arc弧形、拱Infinity无限Composer作曲家AltoTenorBaritoneSopranoRestRhythmTone=pitchScaleChordOrchestra EnsembleBandSolo DuetChoirPlainsongCantataSonataConcertSymphonyConcertoPreludeOvertureOperaMelodyLullabyEuphoniousMovementInstrumentEpisodePercussionWindStringDraftBillEnactRatificationApprovalConfirmationEnforcementDecree=legislation =lawmakingClauseProvisionPrescribeCodificationLegitimateLegalJurisprudenceLegalityContraveneInfringeviolateOffendTransgressAbolish=exterminateAnnulment=abolishCancellation=terminationAbolish=exterminateAnnulment = abolishCancellation = terminationRevocationimmunity豁免权Constitution 宪法Copyright版权、著作Patent专利、执照、专利品Penalty罚款royalties版税Tariff=duty=levy关税Taxation 征税Court法院=tribunalArbitration 调停Delinquency行为不良、违法行为Solicitor 律师attorney律师=lawyerNotaryDefendantProceeding 法律行动Hearing听证会InterrogatoryEvidenceSummonsLiabilityEyewitnessAccusationProsecute = accuse=chargeSueComplaintLawsuitPlea=appealDeposition 革职Indictment = accusation控告Plaintiff起诉Culprit 犯人Recidivist惯犯accomplice同谋者Harbour窝藏Convict罪犯、有罪Acquittal无罪Nonsuit诉讼驳回Verdict 判决= decision =judgmentNonobservance 不遵守、违反adultery通奸Perjury伪证Assassination 暗杀= murder Homicide 杀人Larceny盗窃罪=robbery= theft Swindle = defraudAbduction 诱拐Smuggle走私Embezzlement 侵占、挪用Bribery行贿、受贿Breach = violate 违背Corruption 腐败Slander诽谤Penalty处罚Imprisonment 关押Embargo禁运Indemnity 赔偿、补偿Indemnification 保障、补偿Compensation 补偿Extradition 引渡Domineering 专制的Heirship继承权Confiscate没收Invalidate 作废Captivity 囚禁Trial 审判=hearing= inquisition Detain拘留Extenuate轻微=diminish=lessen Empower 授权Saddle使负担Flee逃跑、逃离Abstinence节制Abstain from eating fat 戒吃肥胖的食物Veto否决V ote投票Rejection 否决Negate 禁止Stipulate 约定、规定Testify证明=verify = give evidenceSubstantiate 证实=corroborate = verifyObservance 遵守Impeach = accuse控告Indictment= charge =accusation起诉Delinquency n. 不法行为;少年罪犯;过失;罪过Incriminate v.控告Denounce 告发Query = inquire质问、询问Exempt = prevent=immune 免除Condone 宽恕=forgive= pardonRemit 赦免Credential 凭证Arable 可耕种的Fertile 肥沃的、富饶的= productiveIrrigate 灌溉Barren 荒地Wasteland 荒地Prairie 大草原、牧场=grasslandPasture 牧地草原Fallow = uncultivated休耕地Stubble 断株Straw稻草Mechanization 机械化、机动化Ranch = farmland大农场Hacienda庄园AgronomistLatifundiumlandlord房东Tenant房客、租用Shepherd 牧羊人vinegrower葡萄栽植者Horticulture n.园艺Dairy 牛奶厂Foodstuff 食品Livestock 牲畜Haystack干草堆Granary n 谷仓Windmill风车Cowshed牛棚Nursery 苗圃Seedbed 苗床Furrow耕Terrace梯田Plantation 种植园Orchard 果园Vineyard葡萄园Tenure 占有Plough 耕地Loosen松土Prune减除Graft嫁接Reclamation开垦、改造Manure施肥FertilizeDungfertilizer肥料SprayInsecticide 杀虫剂Pesticide 杀虫剂Herbicide除草剂Parasite 寄生虫Sickle 镰刀Combine 联合收割机Cereal 谷类Barley 大麦Sorghum高粱植物Prolific 多产的、丰富的Agriculture 农业Aquaculture 水产业Indigenous 土产的Husbandry 耕种Graze吃草Feed吃草Cultivate =till耕种Cultivate= foster= train 培养Hydroponics 水栽培Ridge 起皱Eggplant 茄子Poultry 家禽Buffalo水牛Conservatory 温室GreenhouseSheepfoldPigpen=pigstyTrough 水槽Anthropology人类学Perception感知、感觉Ritual 典礼Stereotype典型Exotic 外来的、奇异的Inexplicable 无法说明的MysteriousTabooPathological 习以为常的Stratification 阶层Tribe 部落Clanethnic种族的Ethnology人种学Minority 少数Descent血统=ancestryHybrid 混血儿Aboriginal 土著=native Ancestor 祖先Forerunner祖先Hominid 原始人类Precursor先驱Predecessor 前辈、前任Antecedent 先辈Racial人种的、种族的Nature 自然、自然界Avalanche 雪崩Mirage 海市蜃楼Innate 天生的、先天的Scenic=picturesque风景优美的Spectacle=sight= scene奇观、景象Jungle丛林Shrub=bush 灌木丛Gorge峡谷Plateau高地、高原Scenery景色Landscape风景Panorama全景Plain平原Tundra冻土地带Iceberg冰山Mountain山Glacier 冰河Deglaciation冰川的小时Valley山谷Peak山顶Range山脉Coast海岸Altitude 高度Watercourse 水道Estuary 河口、江口Matter 物质Vacuum真空Liquid液体、流体Solid 固体Evaporate 蒸发Density 密度Gravity 引力Velocity 速度=speedIntensity 强烈、剧烈Friction摩擦Pressure 压力Vector 向量Temperature 温度Conduction 传导Radiate 射出Expansion 扩充Quantum 量子Dynamics 动力学Inertia惯性mechanics力学Electron 电子Positive 阳的Negative 负电的Charge充电Magnetism吸引力Magnetics 磁学Accelerator 加速者Reflection 反射Mirror镜子Beam光束Image 图像Lens镜片Refraction 折光、折射Focus 焦点Concave 凹面的Convex凸面的Electricity 电流Kinematics 运动学Statics 静力学Magnifier 放大镜Wavelength 波长spectrum光谱Optics光学Optical视力的Translucent透明的Opaque不透明的Transparent=clear = limpid 透明的Current液体、气体流Relativity 相对性Oscillation 摆动Microwave 微波Ultraviolet 紫外线Infrared 红外线的Semiconductor 半导体Insulator 绝缘体Chip芯片Battery 电池Amplifier 扩音器Electromagnetism 电磁Electromagnet 电磁石Acoustic 声学的Sonar 声纳ultrasonics 超声波学Supersonic 超音波的Echo回声Resonance 回声Chafe 擦热=sub= scrape Friction 摩擦Themodynamics 热力学Ventilation 通风= airing = air circulationDeclivity 斜面Dry= dehydrate脱水Dilute = thin = weaken 冲淡Clutter 混乱Chaos混乱= disorder Distillation 蒸馏Centigrade 摄氏的Thermometer 温度计Microscope 显微镜Telescope 望远镜Electron 电子Ion 离子Neutron 中子Nucleus核子Proton 质子atom原子Molecule 分子Emit= discharge = give off 发出、放射Diffuse = scatter = spread传播Vibrate 振动Precipitate 降水Decelerate减速Accelerate 加速Cohesion 结合Elasticity 弹性Compatible 兼容的Harmonious= congruousThaw = melt = defrost 融解Sociology 社会学Hierarchy 层次、等级体系Sociologist 社会学家Marriage 结婚PhenomenonUrban=metropolitanRural = rustic 乡下的Urbanization 都市化Migration 移民、移植Immigration 移居入境Emigration 移民出境Mobility 流动Community 社区Metropolitan大城市的Convention 传统Patriarchic 家长的Institutionalize 习俗Taboo= ban =prohibitionethics伦理学Marital =wedded=conjugal婚姻的polygamous一夫多妻的Tribe = clan 部落Connubial= marital配偶的Matrimony结婚Anathema 令人讨厌的事Linguistics 语言学Language 语言Sound音Dialect 方言Parlance 谈话Intonation 语调、声调Emphatic 用力的、强势的、显然的Paraphrase 改写Syntax 句法Symbol符号Emblem象征Suffix后缀Affix 词缀Prefix 前缀Root词根Sentence 句子Lexical 词汇的Paraphasia 语言错乱的Lingual 语言的Bilingual 能说两种语言的Philology 语言学Semantics 语义学Phonetic 语音的Syllable 音节Parisyllabic 同等音节的Phonemics 音位学Coinage 创造= create = fashion= inventAbridge 缩短= shorten =condense= abbreviateExcerpt = selection = extract 摘录Adaptation 改写Emend修订=amend=improveSynopsis=outline=summaryTag= label=tabGenre=style=mannerSuccinct = terse=conciseGrammarLiteratureGenre类型Satire讽刺文学Fable=legend = allegory=fiction寓言、传说Byword格言Comedy喜剧Tragedy悲剧Masterpiece杰作Author作家Profound=deep深刻的Contemporary当代的Classic杰作Diction措辞、用语Manuscript手稿Analects 文选、论集Literatus学者Renaissance文艺复兴Criticism批评Essay散文、短文Mainstream主流Nostalgia乡愁Connoisseur鉴定家、内行Editorial编辑上的Column专栏Circulation发行量Feature特写Drama戏剧Playwright 剧作家Antenna天线Universe宇宙=galaxy astronomy天文学Planetastronaut宇航员Launch发射Multistage 多级的Cosmos宇宙Sphere球体Celestial=skyGalaxy=nebulapolestar北极星Comet彗星Asteroid小行星Aerolite陨石Satellite人造卫星Constellation星座Nebula星云Equator赤道Zenith顶点Corona日晷MaculaRainbowEclipseOrbit=track=pathClusterJupiterLunarmercurySpaceship=space shuttle Revolve旋转Hawthorn山楂Jasminebotanic植物学Flora植物群Foliage树叶、植物Crossbreed杂种Photosynthesis挂光合作用Peel剥落Shell去壳oyster牡蛎Shoot发芽Vitamin维生素Cell单元、细胞Tissue组织Bud发芽Trunk树干Bark树皮Branch分支Timber木料sprout长出Shrub灌木Fructification果实blossom开花、兴旺、发展Pullulate发芽Tassel穗Zoology动物学Gregarious社交的Fauna动物群Mammmal哺乳动物Carnivore食肉动物Herbiorous食草的Omnivorous杂食的Predator食物动物Prey牺牲者Scavenger清道夫Microbe细菌Reptile爬行动物Primate灵长类的动物Mollusk软体动物Fowl家禽Monster 怪物Herd兽群Swarm蜜蜂群Flock羊群insect昆虫Beast兽Aquatic水生的Amphibian两栖动物Migrate迁徙Graze放牧Peck v.啄WoodpeckerHibernationEstivationDomancyTorporOffspringSpawnPregnantHatchDomesticateFertilizeRegenerationReproduceSqueakChirpCamouflageExtinctionNestNicheHabitatPresidentCongressmanCandidateOpponentAdvocatorAuditorEmperorMonarchDemocratRepublicanRepresentative = delegateAppoint=assignDictator=tyrantAutocrat = ruler=tyrantAristocrat = noblemanRadical= utmostConservativeMinisterGeneral = commanderEnvoy = delegateAmbassador大使consul领事Professor 教授Lecturer 演讲者Dean院长Director主任ChancellorAlumnusPlaywrightMathematicianAstronomerBotanistGeographerGeologistMeteorologistArcheologistArtistArtistArtisan=craftsmaninventor发明家BiographerEcologistCriticConnoisseurCommentator = reviewer = analystCounsel 辩护Lawyer=attorney= Solicitor JuryProxy = agentAgent = proxyDelegate = deputyArbitratorCommanderPrincipalClergyCrew = staffChoreographerEducatorFacultyPilot=aviatorEntrepreneurEmployeeAttendant = waiterStaff = personnelProprietorApprentice = learner = novice= beginner DonorBenefactor =sponsor=patron=supporterBeneficiaryHumanitarian =philanthropistspectator观众Audience听众Resident居民Mortal人类、凡人Believer 信徒Fanatic狂热者Atheist无神论者Adherent 追随者DiscipleZealot 狂热者Assassin 暗杀Burglar = robberBandit = brigand =gangsterbarbarian粗鲁的ExileRebelFoeFigureheadMalcontentSkepticFeministSuperiorInferiorSubordinate =dependantAssistantCoworker考古学Archeology考古学Artifact人造物品Relic遗物Remnant遗迹Skull头脑、骨头Antique古董Antiquity古老遗物Remainsremnant残余RemainderResidueVestigeTraceTrackPrimitivePrehistoricArchaicPrimordialMedievalPrimevaloriginateChronologicalPaleolithicNeolithicExcavationExhumeUnearthScoopPickDigDisclosureInvaluablePreciousCostlyForensicPhysicianPediatricianGynecologistPsychiatristNeurologistDentistSurgeonSanatoriumHygiene=sannitationClinicAilmentIndispositionAffectionUlcerVaccinateChilblainFractureDiagnosisIncubationSymptomSignIndicationRelapseRecur Epidemic Contagion Coma Treatment Anemia Appendicitis Arthritis Bronchities Diabetes Indigestion Influenza Malnutrition Pneumonia Rabies Smallpox Anesthesia Transplant Bandage Acupuncture Contagious Catching Infect Infectious Contract Acute Chronical Morbid Unconscious Fragile Susceptible malady疾病Corpse Tingle Bruise Fester Intoxicate Survive Inject Remedy Prescription Dissect Sterile Clinic Anatomy Sanitation Sustenance Nutrition MalnourishedWelfareV oteVetoRejectionEmbargoSanctionScandalStrikeParadePetitionProcessionIndignityMunicipalIgnominyDomainTerritoryKindomRealmSovereignAutonomySelf-governmentCommissionCommitteeRegimenRegimeElectionBallotIdeologyPrivilegeDispensationAuthorityAuthoritativeSenateCongressFactiousPartisanDiplomaticConfederateConfederacyLeagueAffiliateUnconventionalDictatorialDictatorshipDomesticCentralizeFederalDemocracyConferDiscussEntitleWarrantEmpowerAuthorizeExploitManipulateManeuverContriveConspireInspectScrutinizeExileOstracismHustleImpelOustCoerceBanishEvictDeportDominateAbdicateRelinquishAdministerAdministrationAdministrationInstituteEstablishSet upStratGovernSuperviseManagementInaugurate initiateAmendmentAlterationColonizeMonarchyDominionAnarchismMayhemDoctrineDogmaCreedV olcaniceruption爆发OutburstsquirtMagmaFaultCrustLayer Lithosphere Lithogenous MantleStratum Cataclysm calamityDebacle 泛滥的洪水FloodDelugeMagnitude Seismology Seismic EarthquakeTremorIcebergGlacierDiamondCraterCoreAluminum GeologyPetrifyFossilSedimentOreMineralBonanzaRubyLavaLimestoneGraniteEmerald FieldstoneGemMarblePitBoreholeVeinGeography Ethnography Cosmography GeologyToponymyOceanographyVegetation植被ReliefContinent大陆ArchipelagoPeninsulaIslandMeadowValleyswamp沼泽Lagoon礁湖Moorland 沼泽地Desert 沙漠Dune沙丘Oasis绿洲Savanna热带大草原Tundra苔原Topography 地形学Compass罗盘Meridian 正午Parallel 纬线Longitude 经纬Latitude维度Equator 赤道Zenith 顶点Inlet 小岛Gulf海湾Cliff悬崖Sandbank 沙丘Tempest 暴风雨Seaquake 海啸Estuary入海口Torrent = deluge =floodTributary 支流Confluent 汇合CanalRangeMassifCrevice=cleftPlateauMarineMoistEbbTerrestrialEndemicSubterraneanCavernantarctic南极的antarctica南极洲CoastlandHemisphereContourLowlandNavigationSalinityElevationFormationGeothermicTropicTemperateColossal=immense=hugeProdigious=colossal=enormousMassive =huge=enormousEnormous =huge=vast=immense=tremendousMighty= powerful=strongTremendous = huge=greatImmenseTitanic = huge =immenseGigantic = tremendous =colossalVastMagnitudeMammoth = giantGargantuan = enormousMonstrous = hugeImmeasurableIncalculable = inestimable =innumerableSpaciousExpansiveCommodiousRoomyBroadLooseCapaciousRoomySizeableV oluminousMinute=insighnificant=miniatur eDiminutive = smallMiniature= tinyPetite=slightTrivial= paltryInsignificant = negligible InfinitesimalMinuscule = tinyTrifling = insignificant = trivial Negligible = insighnificant = minimalExcessive= overabundant = inordinateMiscellaneousMultitude = host=mass NumerousRimptionMassivePlentifulAbundant = sufficientCopiousPlenteousAmpleLavishProfuseTeemingProlificSwarmPlethora = superabundance Scarce = sparsePaucityTiny = punyScantSparse = scantyInadequate=deficient=insufficien tScarcityRarenessDearth = scarcity = shortage SlightDeficientUndersuppliedLackingSedate = calm= composed Tranquil = quiet = peaceful = placidStatic = changeless=stagnant Serene=tranquilPlacidStaticSerenePacifySerenityStillLullRestfulQuietudeSqualiditySqualorDinginessFilthyDefileFilthNastinessSoilageDingyMessySlovenlyFrowzyObsceneIndecentImpureBlemishSmearSoptSmirchAttaintTaintStainGrimyBegrimedBlotchSmudgeSplotchRemotedistant=inaccessibleDistant=remoteProximity = nearnessAdjacent = adjoining= neighboringAdjoin=abutAdjoining = adjacent= neighboringPropinquityNeighboringBorderingAbutCircumjacentBorderAcid=sour=tartSavor=taste = relishAura = aromaSmellFlavorPungentFragrant = aromaticPalatableBalmyDeliciousSourPerfumeBitterOdorousVinegaryPepperyFishyGloriousGorgeous=beautiful=admirable=colorfulGrand=splendid = magnificentMagnificent = gallant = splendidRadiant = joyous = beamingSolemn = grave = somberSpectacular = breathtaking =impressive = strikingSplendid = magnificentSplendor = grandeur =magnificenceSuperb = excellent = first-rateCardinal = essentialChiefly = mainly = principallyCrucial = decisive = criticalDominant = predominant =prevalentElementaryElite = bestEssential = crucial = vital =fundamental = necessaryFatefulForemost = primeForte = strong pointFundamental = essential =elementaryGist = themeKey = dorminant = primary Largely = mainly = for the most partLeading = principal = chief MajorMerit = advantageDemeritMomentous = important = criticalMotif = theme = subject Optimum = bestPivotalPredominantly = primarily = chiefly = principally Predominate = prevail Preference = inclination = privilegePrimarily = chiefly = mainly = principallyPrincipal = main = chief = centralSignificant = importantSole = only = mere= exclusive SumSuperbTenor = natureVital = important = crucial = essentialActual = practical = real Authentic = genuine = real GenuineTangible = touchable = substantialTruly = genuinely = actually Unfeigned = sincere VeraciousMendaciousVeridicalConceive = devide = visualize Fantasy = dream = fancy Fictitious = invented = imaginaryFigmentIllusion = hallucination ImaginaryImaginative Mythical = legendary= fictiousOccult = supernaturalSuperstitionVisionalLegendaryDelusionHallucinationMysteriousPanaceaCure-allOmnipotentAmightyAll-powerfulSupremeElixirNostrumCompact = packedCrisp = crunchy =brittleCrooked = bent =twistedDense = thick = closeDepressionEthereal = light =delicateHectic = feverishImpervious =impenetrableLithe = flexible =suppleLofty = high =toweringMalleable = pliablePonderous = heavyShallow = superficialSlim = slender = thinSlippery = slick =smoothSloppySuppleTenuous = thin =weakCriticalDeleterious = harmful= detrimentalDeterment =determentalEndangerHarmHarmfulHazardHazadous = dangerous =perilousImpair = harm = damageJeopardy = danger = riskMaim = diable = mutilatePerilPerilous = dangerous =hazardousThreaten = menace = terrifyCrisis = turning point =emergencyGrave = serious = significantRuinous = destructiveImperil = endangerVulnerabilityPrejudiceCharacter性格Characteristic 特点、特有的Extraordinary = remarkable =outstandingGiven = specifiedIdiosyncrasyOriginalParticularly = especiallyQuaint = queer = oddTrait = characteristic = attributeAttribute = characteristic =featureUniqueAbnormal = exceptionalBizarre = odd = erratic =eccentricErraticQueerWeirdOddityWhimsicalAberrant = deviatingAdequate = sufficientBlockBulk = mass = volume =majorityCountless = innumerableExcess = surplus。
专利名称:用于鉴别调节诺里蛋白、诺里蛋白模拟物的药剂的物质和方法以及由此鉴别出的药剂
专利类型:发明专利
发明人:弗雷德里克·J·贝克斯三世,比姆·M·巴特
申请号:CN200780002066.2
申请日:20070206
公开号:CN101365945A
公开日:
20090211
专利内容由知识产权出版社提供
摘要:本说明书揭示用于筛检和鉴别调节与Wnt通路信号转导有关的诺里蛋白活性的试剂的物质和方法。
由此鉴别出的药剂优选调节骨重塑和/或脂质含量,且可为诺里蛋白模拟物和诺里蛋白激动剂以及LRP5/诺里蛋白/卷曲蛋白4复合物的其它激动剂和模拟物。
申请人:惠氏公司
地址:美国新泽西州
国籍:US
代理机构:北京律盟知识产权代理有限责任公司
代理人:刘国伟
更多信息请下载全文后查看。
Okay, class! Today were are going to discuss Ragtime music. It’s a genre that some of you may have already heard of, but will likely be new for most of you. First, remember that a genre is a type of classification. For example, Hip-Hop and Rock are genre s of music, and you should be able to think on your own how each ofthose genre s comes up with a set of characteristics that help define each type of music and set it apart from other genre s. To continue with our previous examples, Hip-hop often features rap and a particular emphasis on the beats. Rock, on the other hand, refers to music that mostly includes a four-piece band: 2 guitarists, 1 bass-player, and 1 drummer. So then, what characterize s the genre known as Ragtime? Well, Ragtime music is mostly defined by the presence of a shifted or ragged rhythm. This means that the regular flow or rhythm of a piece of music will be disrupted by a shift, making it sound slightly off-beat. Ragtime music is considered to be lively and springy, which makes it a great genre for dancing to.Let’s look more specifically at the form of Ragtime music. Ragging, or creating a ragged rhythm, was first a modification of the march made popular by John Philip Sousa. Additional poly rhythm s then came from African music. Ragtime itself was usually written as 2/4 or 4/4 time with a predominant left-hand pattern of bass notes on strong beats and chords on weak beats, accompanied by a modified melody in the right-hand. According to some sources, the name ragtime may very well have come from the ragged rhythm of the right hand. There are also rags written in 3/4 time, which are known specifically as a ragtime waltz.And while we may be focusing on the time signatures that Rag is written in, I want you to all understand something very important though: Ragtime is not strictly confined to any one time signature, like how a march will always be in 2/4 and a waltz will always be in 3/4. Instead Ragtime is…umm...it's..a...a musical genre that can be applied to any meter…so rather than being one specific ti me signature, it can take any time signature and syncopates the beat to become Ragtime. The defining characteristic of Ragtime music is that type of syncopation in which melodic accents occur between metrical beats. This results in a melody that seems like it avoids some metrical beats of the accompaniment by emphasizing notes that either anticipate or follow the beat. Now this might sound a bit complicated, but the ultimate intention is to accentuate the beat and get the listener to move to the music.Now that you know a little about what Ragtime is, let’s discuss its history. Ragtime was popular between the 1890’s and the early 1900’s. It began as dance music in the red-light districts of African American communities in St. Louis and New Orleans. It was many years before Ragtime actually became published as popular sheet music for piano. The man who is credited for developing this genre of music is Ernest Hogan. Hogan began working as a comedian and entertainer, producing many shows that featured the syncopation that became integral to the Ragtime genre. Movingon….Scott Joplin is also a notable figure in the history of Ragtime. He becamefamous for composing “Maple Leaf Rag” in 1899 and then a string of ragtime hits, such as “The Entertainer”. I’d be willing to bet that mos t of you today even know “The Entertainer” or have heard it at some point in your lives. Anyway…one of the key take away here is that Scott Joplin is easily the biggest ragtime composer and his music went on to influence many other composers for a number of years.Alas, as we move forward in time, Ragtime eventually declined in popularity and jazz began to claim the public’s attention. In 1917, Ragtime was, more or less, officially pushed out by Jazz. That being said, jazz likely wouldn’t have happened without Ragtime. Just like Ragtime built off the marches which were popular before it, jazz built off of ragtime as well.To cap off this discussion, I want to end with a quote from Scott Joplin. Known as the King of Ragtime, Joplin called the effects of the music “weird and intoxicating”. He also used the term swing to describe how to play ragtime and would say “play it slowly until you catch the swing”. So, take a listen to so me Ragtime music, and let yourself find the swing in it. While Ragtime may not be your cup of tea, just make sure you understand its importance along the wide spectrum of music and that music is a universal language that speaks to us all.classificationA category into which something is put.Sentence UsageSeveral classification s are used to categorize individual apple treesgenreA category of artistic composition, as in music or literature, characterized by similarities in form, style, or subject matter.Sentence UsageWomen also bring to poetry or other genre s of literature a whole new area of experience and vision.characterizebe typical or characteristic ofSentence UsageThe disease is characterize d by weakening of the immune system. It was a social relationship characterize d by an unequal distribution of power and resources. predominantPresent as the strongest or main element:;Having or exerting control or power: Sentence UsageThe three predominant colours of this film are black, white and green. In 1914, France was one of Europe's leading powers but not the predominant force that it had been in Napoleon's day.accentuateMake more noticeable or prominentSentence UsageInstead of focusing on the length of your cut, concentrate on accentuating your best feature.rhythmA strong, regular, repeated pattern of movement or sound:Sentence UsageWhether it's in the form of romantic melody, upbeat Swing Jazz or exotic world rhythm s, the live musical experience adds a unique presence and excitement to any event.intoxicatingExhilarating or exciting;(Of alcoholic drink or a drug) liable to cause intoxication. Sentence UsageRefrain from intoxicating drink and drugs which lead to carelessness.。
Aameloblast 成釉细胞amelogenesis 釉质形成amelogenins 釉原蛋白non-amelogenins 非釉原蛋白ameloblastin 成釉蛋白abrasion/attrition 磨损acellular cementum 无细胞牙骨质attached gingiva 附着龈apoptosis 细胞凋亡alveolar bone/process 牙槽骨/窝alveolar bone proper/lamina dura 固有牙槽骨/硬骨板attachment plaque 附着斑anchoring fibril 锚纤维acinus(serous/ mucous/mixed浆液/粘液/混合性腺泡Bbranchial arch 鳃弓bud stage 蕾状期bell stage 钟状期basement membrane (zone)基底膜(区)Bartholin duct 舌下腺主导管Cconotruncal 主动脉的cleft lip 唇裂cleft palate 腭裂cleft jaw 颌裂copula 联合突(舌的发育)cap stage 帽状期/增殖期cervical loop 颈环(外釉上皮和内釉上皮相连处)cariostatic potential 耐龋潜能cross striations 横纹(垂直釉柱,每天速度)circumpupal dentin 髓周牙本质cementum 牙骨质cementoid 类牙骨质cellular cementum 细胞牙骨质cornified envelope 角化包膜DDiGeorge syndromedental lamina 牙板dental papilla 牙乳头dental sac 牙囊DPP(dentin phosphoproteins)牙本质磷蛋白dental tubule 牙本质小管dead tract 死区direct innervation theory 神经传导学说(牙本质感觉)desmosome 桥粒dendtritic cells 树枝状细胞DCJ dentino-cemental junction 牙本质牙骨质界dentogingival junction 牙龈结合demilune 半月板dense body致密小体Eectomesenchyme 外胚间叶、外间充质enamel 釉质enamel organ 成釉器enamel knot 釉结(牙胚中央,内釉上皮局部增厚,调节细胞分化和牙形态发生的信号中心)enamel cord 釉索(釉结想外釉上皮走行的细胞条索,成釉器一分为二,参与决定牙尖的早期位置)enamel niche 釉龛(片状牙板内凹成的腔隙,内满结缔组织enamelin 釉蛋白enamel rod 釉柱enamel rod sheath 釉柱鞘(晶体长轴、釉柱长轴65-70釉柱间隙头清晰弧形边界)enamel spindle 釉梭enamel tufts 釉丛enamel lamellae 釉板enamel cuticle 釉小皮EDJ enamel-dentinal junction 釉质牙本质界ECJ enamel-cementum junction 釉质牙骨质界eruption 萌出elastin fibers(Oxytalan /Eluanin弹性纤维epithelial ridges 上皮嵴exocytosis 胞吐excretory duct 排泄管Ffrontonasal process 额鼻突fuse 融合(突起之间外胚层相互接触、破裂、退化、消失)facial cleft 面裂foramen cecum 舌盲孔free gingiva 游离龈free gingival groove 游离龈沟filiform papilla 丝状乳头fungiform papilla 菌状乳头foliate papilla 叶状乳头Gganglionic placode 神经节原基globular process 球状突glycosaminoglycans 氨基葡萄糖gubernacular canal 引导管gnarled enamel 绞釉gap junction 缝隙连接gingiva 牙龈gingival sulcus 龈沟gingival col 龈谷gingival epithelium 牙龈上皮goblet cell 杯状细胞Hholoprosencephaly 前脑单脑室畸形Hertwing 上皮根鞘Hertwig’s epithelial root sheath 上皮根鞘(牙根发育时,内釉外釉上皮在颈环处曾生成双层)hyaline layer 透明层hydrodynamic theory 流体动力学说histiocyte and undifferentiated mesenchymal cell 组织细胞和非分化间充质细胞Howship 陷窝Iincisive canal/naso-platal canal切牙管/鼻腭管inner enamal epithelium 内釉上皮层incremental lines ( lines of Retzius 釉质生长线Von Ebner 牙本质生长线intertulubar dentin 管间牙本质interglobular dentin 球间牙本质intermediate junction 中间连接intermediate cementum 中间牙骨质interdental papilla 牙间乳头intercalated duct 闰管Jjunctional complex 细胞连接复合体junctional epithelium 结合上皮K KHN(Knoop hardness number)洛氏硬度值Kroff 纤维Llateral nasal process 侧鼻突lateral palatal process/second plate 侧腭突/继发腭lateral lingual prominence/swelling 侧舌隆突lamina limitans 限制板lamina lucida 透明板lamina densa 密板lamina reticularis 网板lamina propria 固有层lining mucosa 被覆粘膜lingual follicle 舌滤泡lingual crypt 舌隐窝Langerhans cell 朗格汉斯细胞Mmaxillary process 上颌突medial nasal process 中鼻突merge 联合(面部发育时突起之间的沟变浅、消失)Meckel cartilage 第一鳃弓软骨/下颌软骨mantle dentin 罩牙本质(牙冠的原发性牙本质,纤维粗大)matrix vesicle 基质小泡Malassez epithelial rest马拉瑟上皮剩余/牙周上皮剩余(牙骨质形成时,剩余上皮细胞离开牙根表明,保留在发育的牙周膜中,牙根发育期上皮根鞘残留)melanocyte 黑色素细胞Merkel cell 梅克尔细胞masticatory mucosa 咀嚼粘膜myoepithelial cell/basket cell肌上皮细胞/篮细胞myofilament 肌微丝Nnasal pit 鼻凹/嗅窝nasal fin 鼻鳍nanospheres 纳米球neonatal line 新生线Neumann sheath 诺依曼鞘Oorapharyngeal membrane 口咽膜olfactory placode/nasal placode 嗅板/鼻板outer enamal epithelium 外釉上皮层odontoblast (secretory/resting)成牙本质细胞(分泌型/静止型)odontoblastic process 成牙本质细胞突起Owen line 欧文线oral mucosa 口腔粘膜orthokeratinization 正角化oncocytes 嗜酸细胞oncocytoma 嗜酸粒细胞瘤oxiphilic adenoma嗜酸性腺瘤Ppatterning 模式发育primary epithelial band 原发性上皮带preodontoblast 前成牙本质细胞predentin 前期牙本质(成牙本质细胞层与矿化牙本质之间的有机基质)proteinases 蛋白酶perikymata 釉面横纹/牙面平行线pulpo-dentinal complex 牙髓牙本质复合体preiodontoblastic space 成牙本质细胞突周间隙peritubular dentin/intratubular dentin 管周牙本质/管内牙本质primary dentin 原发性牙本质pulp 牙髓pulp proper/core 固有牙髓/髓核periodontal membrance/ligamant 牙周膜/韧带parakeratinization 不全角化parotid gland 腮腺QRRAS retinoic acid syndrom 维甲酸综合征Reicher软骨第二鳃弓软骨Rathke pouch 拉特克囊reduced dental epithelium 缩余釉上皮(釉质发育完成后,成、中、星、外结合形成鳞状上皮盖在釉小皮上)rodless enamel 无釉柱釉质reparative/teritiary/reaction dentin/osteodentin 修复性/第三期/反应性/骨样牙本质Raschkow 丛(parietal layer of nerves)神经壁层reversal line 反转线Ruffini 末梢(根尖周围牙周膜神经纤维)reserve cell 储备细胞Sstomadeum/oral pit 原口/口凹sulcus terminalis 界沟stellate reticulum 星网状层stratum intermedium 中间层(内釉和星网状层之间)Serre 上皮剩余(残留的牙板上皮,以上皮岛/团在颌骨、牙龈中)shedding (乳恒牙)交替Schreger line 施雷格线secondary dentin 继发性牙本质Sharpey/perforating fiber 沙比/穿通纤维stratum basale/germinativum 基底层/生发层stratum spinosum 棘层stratum granulosum 颗粒层stratum corneum 角化层submucosa 粘膜下层specialized mucosa 特殊粘膜saliva 唾液salivary glands唾液腺secretory unit 分泌单位seromucous cells 浆粘液细胞secretory duct 分泌管striated duct 纹管stensen duct 腮腺导管submandibular gland 颌下腺sublingual gland 舌下腺TTreacher Collin syndrometuberculum impar 奇结节thyroglossal duct 甲状舌管terminal web 终棒Tomes processes 成釉细胞突/托姆斯突tuftelin 釉丛蛋白tropocollagen 原胶原Tomes granular layer 托姆斯颗粒层transparent/sclerotic dentin 透明/硬化性牙本质transduction theory 转导学说(牙本质感觉)tight junction 紧密连接taste bud 味蕾Vvallate papilla 轮廓乳头von Ebner gland 味腺Wwedge shaped defectWeil 层(牙髓无细胞层)Warton duct 颌下腺主导管。
超声引导下周围神经阻滞技术新进展金荒漠;郭向阳【摘要】超声引导下周围神经阻滞技术目前得到了广泛的认可.与单纯使用神经刺激器相比,超声引导下神经阻滞的成功率更高,耗时更短.超声引导神经阻滞技术近年来得到了较快的发展,但将超声应用于硬膜外麻醉是否可行仍存在争议.超声引导技术并不能显著减少穿刺所致神经损伤的发生率,但与神经刺激器联合使用仍能改善神经阻滞操作的安全性.【期刊名称】《中国继续医学教育》【年(卷),期】2011(003)010【总页数】8页(P51-58)【关键词】超声;周围神经阻滞;神经刺激器;局部麻醉;神经损伤【作者】金荒漠;郭向阳【作者单位】北京大学第三医院麻醉科;北京大学第三医院麻醉科【正文语种】中文安全、便捷、有效和快速的医疗工作是现代医学的核心内容,麻醉科的许多临床工作针对的是病人的神经系统和心血管系统,怎样在高效运转的医疗模式下保证医疗质量,是摆在麻醉学从业人员面前的严峻问题。
目前,以超声技术为代表的诸多可视化技术使麻醉学科逐渐摆脱了“盲目”操作的时代,进入到可视化操作的新纪元。
超声技术是近年来发展最快、认可度最高的可视化技术。
它对医疗效率和医疗质量的提高起到了极大的推动作用,改善了临床麻醉操作的水平,将麻醉学带入了一个崭新的时代。
麻醉超声技术包括超声引导神经阻滞技术、经食道超声心动图技术(TEE)、经颅多普勒技术、以及超声引导下的动静脉穿刺等等。
其中超声引导周围神经阻滞技术近年来以令人惊讶的速度得到了人们的认可,值得我们在下面进一步阐述。
1 超声引导技术与应用神经刺激器在外周神经阻滞研究进展周围神经阻滞技术自发明以来已经有多种辅助方法,如超声引导[1]、透视引导[2]、神经刺激[3]和筋膜突破音[4]等,从最初的寻找异感法,到神经刺激器的广泛使用,再到超声引导辅助神经刺激器,周围神经阻滞技术的安全性和有效性经历了逐步提高的发展过程。
近年来有多项研究对神经刺激器或超声引导技术进行了比较,有证据显示,单纯通过神经刺激方法引导置管对改善下肢的镇痛效果作用很小[5-6]。
口腔正畸学单词Orthodontics 口腔正畸学American Board Orthodontists 美国正畸学会Differentiation 分化Translocation 改位、易位Orthopedic devices 矫形治疗措施Functional jaw orthopedic 功能颌骨矫形Cephalocaudal gradient of growth生长的头尾增减率Pattern of growth 生长型Pattern of facial growth 面部生长型Average growth pattern 平均生长型Horizontal growth pattern 水平生长型Vertical growth pattern 垂直生长型Balanced growth 平衡生长Growth variability 生长变异Chronometry 颅测量术Cephalometic radiography 线头影测量术Displacement 骨移位Primary displacement 原发性骨移位Secondary displacement 继发性骨移位Premature hypostasis 骨缝早融Skeletal craniofacial developmentsyndrones颅面发育综合征Oxycephaly 尖头畸形Brachycephaly 短头畸形Scaphocephaly 舟状头畸形Chondrocranium 软骨性颅Desmocranium 头颅Cranial base 颅底Spheno-occipital synchondrosis 蝶枕软骨联合Inter-sphenoid synchondrosis蝶骨间软骨联合Spheno-ethmoid synchondrosis蝶筛软骨联合Growth of masomaxillary complex鼻上颌复合体的生长Masomaxillary complex 鼻上颌复合体Nasal septum 鼻中隔Frontal-maxillarysuture 额颌缝Eygomatic-maxillarysurure 颧颌缝Eygomatic-temporal suture 颧颞缝Pterygo-palatin suture 翼颌缝Mandible下颌骨Genial angle 下颌角External rotation 外旋转Internal rotation 内旋转T otal rotation 总旋转Intramatrix rotation 基质内旋转Hypertrophy 肥大Hyperplasia 增生Acceptable compromises 可接受的折中值T reatment goal individualized个体化治疗目标T otal discrepancy 上下牙弓总不调量Discrepancy 牙弓不调量Available space 可用间隙Requried space 必需间隙Relocation 复位Expansion 扩大牙弓Intermaxillary 颌间Extraction 拔牙supra- 表示[在上; 在远方]之义Visual treatment objective 治疗目标预测法Mesiofacial type 中间型Brachyfacial type 短面型Dalichofacial type 长面型Guadrilateral analysis 四边形分析法Sassauni analysis 正位片(后前位)的分析法Ritucci-Burston 颏顶位分析法Soft tissue facial angle 软组织面角Nose prominence 鼻突度Superior sulcus depth 上唇沟深H-line to subnasale 鼻下点至H线距Skeletal profile convexity 骨侧面突度Upper lip strain measurement 上唇紧张度Lower lip to H-line 下唇H线距Inferior sulcus to H-line 颏唇沟深度Soft tissue chin thickness 颏部软组织厚度Esthetic plane、E-Plane 审美平面Mesioversion 近中错位Distoversion 远中错位Linguoversion 舌向错位Labioversion 唇向错位Infraversion 低位(牙合下错位)Supraversion 高位(牙合上错位)T orsiversion 旋转Axisversion 斜轴Transversion 易位Ba-N 全颅底平面N-Pog 面平面Nba-PtGn 面轴角Pt-Gn 面轴FH-Npog 面角FH-MP 下颌平面角MP-NPg 颏角ANS-Xi-Pm 下面高角Dc-Xi-Pm 下颌弓角A-Npog A点突度L1-Apog 下中切牙突距,下中切牙倾斜度PTV-U1 上颌第一磨牙位置L1-EP 下唇位置N 鼻根点S 蝶鞍点Ba 颅底点Bo Bolton点Po 耳点Or 眶点ANS 前鼻棘点PNS 后鼻棘点Ptm 翼上颌裂点Pt 翼点B 下牙槽座点Pm Pm点,下颌前缘部B点到颏前点间,由凹至凸的移行交界点。
一种多特征融合的说话人辨认算法作者:孙佳宁于玲来源:《电脑知识与技术》2022年第15期摘要:针对在智能音箱中容易出现误唤醒情况,即设备被环境音错误激活的问题,该文提出了一种多特征融合的说话人辨认算法。
该算法在特征提取部分通过将短时能量、线性预测倒谱系数(LPCC)、梅尔频率倒谱系数(MFCC)及其一阶动态特征差分系数进行有机结合来提高说话人辨认算法的识别率。
使用自建语音库进行仿真测试,仿真实验结果表明,与采用传统特征提取的GMM说话人辨认相比,采用改进的特征提取方法能显著提高说话人辨认的识别正确率。
关键词:说话人辨认;MFCC;LPCC;短时能量中图分类号:TP18 文献标识码:A文章编号:1009-3044(2022)15-0082-03在实际生活中,智能音箱容易出现误唤醒的情况,比如电视里提到唤醒词,或者外面小朋友贪玩喊出唤醒词,都会导致误唤醒的发生。
在进行唤醒词识别前,加入对说话人的辨认[1]可以有效减少这种情况的发生。
说话人辨认的性能主要取决于特征提取和模式识别部分。
目前常用的特征有梅尔频率倒谱、感知线性预测、线性预测倒谱[2]。
采用单一的线性预测倒谱特征(LPCC)对语音的清音识别来说并不准确;采用单一的短时能量可以准确区分清浊音,但抗噪性很差;采用单一的梅尔频率倒谱特征(MFCC)抗噪性比较强,但其各维分量对识别性能的贡献是不同的,如第一维、第二维特征分量会使说话人辨认的识别效果更差[3]。
故本文考虑通过多特征融合来提高说话人辨认的识别准确率,进而降低智能音箱的误唤醒率。
本文在特征提取部分,将LPCC、MFCC及其一阶动态特征差分系数进行有机结合,并将MFCC中第一维特征分量舍弃并替换为短时能量,获取说话人特征的更多信息,从而有效提高说话人辨认系统的识别性能。
与采用单一特征进行说话人辨认相比,多特征融合的说话人辨认算法抗噪声性能更强,在环境适应性方面更有优势。
1说话人辨认的特征提取本文所采用的特征包括:线性预测倒谱系数(LPCC)、短时能量、梅尔频率倒谱系数(MFCC)及其一阶动态特征。
Classification-Based Melody TranscriptionDaniel P.W.Ellis and Graham E.PolinerLabROSA,Dept.of Electrical EngineeringColumbia University,New York NY10027USA{dpwe,graham}@February15,2006AbstractThe melody of a musical piece–informally,the part you would hum along with–is a useful and compact summary of a full audio recording.The extractionof melodic content has practical applications ranging from content-based audioretrieval to the analysis of musical structure.Whereas previous systems generatetranscriptions based on a model of the harmonic(or periodic)structure of musicalpitches,we present a classification-based system for performing automatic melodytranscription that makes no assumptions beyond what is learned from its trainingdata.We evaluate the success of our algorithm by predicting the melody of theADC2004Melody Competition evaluation set,and we show that a simple frame-level note classifier,temporally smoothed by post processing with a hidden Markovmodel,produces results comparable to state of the art model-based transcriptionsystems.1IntroductionMelody provides a concise and natural description of music.Even for complex,poly-phonic signals,the perceived predominant melody is the most convenient and memo-rable description,and can be used as an intuitive basis for communication and retrieval e.g.through query-by-humming[e.g.Birmingham et al.,2001].However,to deploy large-scale music organization and retrieval systems based on melodic content,we need mechanisms to automatically extract the melody from recorded audio.Such transcrip-tion would also be valuable in musicological analysis as well as numerous potential signal transformation applications.Although automatic transcription of polyphonic music recordings(multiple instru-ments playing together)has long been a research goal,it has remained elusive.In Goto and Hayamizu[1999],the authors proposed a reduced problem of transcribing polyphonic audio into a single melody line(along with a low-frequency bass line)as a useful but tractable analysis of audio.Since then,a significant amount of research has taken place in the area of predominant melody detection,including Goto[2004], Eggink and Brown[2004],Marolt[2004],Paiva et al.[2004],Li and Wang[2005].1These methods,however,all rely on a core analysis that assumes a specific audio structure,specifically that musical pitch is produced by periodicity at a particular fun-damental frequency in the audio signal.For instance,the system of Goto and Hayamizu [1999]extracts instantaneous frequencies from spectral peaks in the short-time analy-sis frames of the music audio;Expectation-Maximization is used tofind the most likely fundamental frequency value for a parametric model that can accommodate different spectra,but constrains all the frequency peaks to be close to integer multiples(harmon-ics)of the fundamental,which is then taken as the predominant pitch.This assumption that pitch arises from harmonic components is strongly grounded in musical acoustics, but it is not necessary for transcription.In manyfields(such as automatic speech recog-nition)classifiers for particular events are built without any prior,explicit knowledge of how they are represented in the features.In this paper,we pursue this insight by investigating a machine learning approach to automatic melody transcription.We propose a system that infers the correct melody label based only on training with labeled examples.Our algorithm identifies a single dominant pitch,assumed to be the melody note,for all time frames in which the melody is judged to be sounding.The note is identified via a Support Vector Machine classifier trained directly from audio feature data,and the overall melody sequence is smoothed via a hidden Markov model,to reflect the temporal consistency of actual melodies. This learning-based approach to extracting pitches stands in stark contrast to previous approaches that all incorporate prior assumptions of harmonic or periodic structure in the acoustic waveform.The main contribution of this paper is to demonstrate the feasibility of this approach,as well as describing our solutions to various consequent issues such as selecting and preparing appropriate training data.Melody,or predominant pitch,extraction is an attractive task for the reasons out-lined above,but melody does not have a rigorous definition.We are principally in-terested in popular music which usually involves a lead vocal part–the singing in a song.Generally,the lead vocal will carry a melody,and listeners will recognize a piece they know based only on that melody alone.Similarly in jazz and popular instrumental music,there is frequently a particular instrument playing the lead.By this definition, there is not always a melody present–there are likely to be gaps between melody phrases where only accompaniment will be playing,and where the appropriate output of a melody transcriber would be nothing(unvoiced);we will refer to this problem as voicing detection,to distinguish it from the pitch detection problem for the frames containing melody.There will always be ambiguous cases–music in which several instruments contend to be considered the‘lead’,or notes which might be considered melody or perhaps are just accompaniment.Of course,a classification system simply generalizes from the training data,so to some extent we can crystallize our definition of melody as we create our training ground truth.On the whole,we have tended to stay away from borderline cases when selecting training data(and the standard test data we use has been chosen with similar criteria),but this ambiguity remains a lurking prob-lem.For the purposes of this paper,however,we evaluate melody transcription as the ability to label music audio with either a pitch–a MIDI note number,or integer corre-sponding to one semitone(one key on the piano)–or an“unvoiced”label.Resolving pitch below the level of musical semitones has limited musical relevance and wouldfit less cleanly into the classification framework.2The remainder of this paper is structured as follows:Training data is the single greatest influence on any classifier,but since this approach to transcription is unprece-dented,we were obliged to prepare our own data,as described in the next section.Sec-tion3then describes the acoustic features and normalization we use for classification. In section4we describe and compare different frame-level pitch classifiers we tried based on Support Vector Machines(SVMs),and section5describes the separate prob-lem of distinguishing voiced(melody)and unvoiced(accompaniment)frames.Section 6describes the addition of temporal constraints with hidden Markov models(HMMs) by a couple of approaches.Finally,section7discusses the results,and presents some ideas for future developments and applications of the approach.2Audio DataSupervised training of a classifier requires a corpus of labeled feature vectors.In gen-eral,greater quantities and variety of training data will give rise to more accurate and successful classifiers.In the classification-based approach to transcription,then,the biggest problem becomes collecting suitable training data.Although the availability of digital scores aligned to real recordings is very limited,there are some other possible sources for suitable data.We investigated using multi-track recordings and MIDI audio files for training;for evaluation,we were able to use some recently-developed standard test sets.2.1Multi-track RecordingsPopular music recordings are typically created by layering a number of independently-recorded audio tracks.In some cases,artists(or their record companies)make available separate vocal and instrumental tracks as part of a CD or12”vinyl single release. The‘acapella’vocal recordings can be used to create ground truth for the melody in the full ensemble music,since solo voice can usually be tracked at high accuracy by standard pitch tracking systems[Talkin,1995,de Cheveigne and Kawahara,2002].As long as we can identify the temporal alignment between the solo track and the full recording(melody plus accompaniment),we can construct the ground truth.Note that the acapella recordings are only used to generate ground truth;the classifier is not trained on isolated voices since we do not expect to use it on such data.A collection of multi-track recordings was obtained from genres such as jazz,pop, R&B,and rock.The digital recordings were read from CD,then downsampled into monauralfiles at a sampling rate of8kHz.The12”vinyl recordings were converted from analog to digital monofiles at a sampling rate of8kHz.For each song,the fundamental frequency of the melody track was estimated using fundamental frequency estimator in WaveSurfer,which is derived from ESPS’s get f0[Sj¨o lander and Beskow, 2000].Fundamental frequency predictions were calculated at frame intervals of10ms and limited to the range70–1500Hz.We used Dynamic Time Warping(DTW)to align the acapella recordings and the full ensemble recordings,along the lines of the procedure described in Turetsky and Ellis[2003].This time alignment was smoothed and linearly interpolated to achieve a3f r e q / H z f r e q / H z500100015002000-5R e l a t i v e M e l o d i cP o w e r (d B )time / sec102030Figure 1:Examples from training data generation.The fundamental frequency of the isolated melody track (top pane)is estimated and time-aligned to the complete audio mix (center).The fundamental frequency estimates,rounded to the nearest semitone are used as target class labels (overlaid on the spectrogram).The bottom panel shows the power of the melody voice relative to the total power of the mix (in dB);if the mix consisted only of the voice,this would be 0dB.frame-by-frame correspondence.The alignments were manually verified and corrected in order to ensure the integrity of the training data.Target labels were assigned by calculating the closest MIDI note number to the monophonic prediction.Example signals from the training data are illustrated in figure 1.We ended up with 12training excerpts of this kind,ranging in duration from 20s to 48s.Only the voiced portions were used for training (we did not attempt to include an ‘unvoiced’class at this stage),resulting in 226s of training audio,or 22,600frames at a 10ms frame rate.2.2MIDI AudioMIDI was created by the manufacturers of electronic musical instruments as a digital representation of the notes,times,and other control information required to synthesize a piece of music.As such,a MIDI file amounts to a digital music score that can easily be converted into an audio rendition.Extensive collections of MIDI files exist consisting of numerous transcriptions from eclectic genres.Our MIDI training data is composed of several frequently downloaded pop songs from www.fi.The4trainingfiles were converted from the standard MIDIfile format to monaural audiofiles (.W A V)with a sampling rate of8kHz using the MIDI synthesizer in Apple’s iTunes. Although completely synthesized(with the lead vocal line often assigned to a wind or brass voice),the resulting audio is quite rich,with a broad range of instrument timbres, and including production effects such as reverberation.In order to identify the corresponding ground truth,the MIDIfiles were parsed into data structures containing the relevant audio information(i.e.tracks,channels numbers,note events,etc).The melody was isolated and extracted by exploiting MIDI conventions:Commonly,the lead voice in pop MIDIfiles is stored in a monophonic track on an isolated channel.In the case of multiple simultaneous notes in the lead track,the melody was assumed to be the highest note present.Target labels were determined by sampling the MIDI transcript at the precise times corresponding to the analysis frames of the synthesized audio.We usedfive MIDI excerpts for training,each around30s in length.After removing the unvoiced frames,this left125s of training audio(12,500frames).2.3Resampled AudioIn the case when the availability of a representative training set is limited,the quantity and diversity of musical training data may be extended by re-sampling the recordings to effect a global pitch shift.The multi-track and MIDI recordings were re-sampled at rates corresponding to symmetric semitone frequency shifts over the chromatic scale (i.e.±1,2,...6semitones);the expanded training set consisted of all transpositions pooled together.The ground truth labels were shifted accordingly and linearly inter-polated to maintain time alignment(because higher-pitched transpositions also acquire a faster tempo).Using this approach,we created a smoother distribution of the train-ing labels and reduced bias toward the specific pitches present in the training set.Our classification approach relies on learning separate decision boundaries for each indi-vidual melody note with no direct mechanism to ensure consistency between similar note classes(e.g.C4and C#4),or to improve the generalization of one note-class by analogy with its neighbors in ing a transposition-expanded training restores some of the advantages we might expect from a more complex scheme for tying the parameters of pitchwise-adjacent notes:although the parameters for each classifier are separate,classifiers for notes that are similar in pitch have been trained on transposi-tions of many of the same original data frames.Resampling expanded our total training pool by a factor of13to around456,000frames.2.4Validation and Test SetsResearch progress benefits when a community agrees a consistent definition of their problem of interest,then goes on to define and assemble standard tests and data sets. Recently,the Music Information Retrieval(MIR)community has begun formal evalua-tions,starting with the Audio Description Contest at the2004International Symposium on Music Information Retrieval(ISMIR/ADC2004)[Gomez et al.,2004].Its succes-sor is the Music Information Retrieval Evaluation eXchange(MIREX2005)[Downie5et al.,2005].Each consisted of numerous MIR-related evaluations including predomi-nant melody extraction.The ADC2004test set is composed of20excerpts,four from each offive styles,each lasting10-25s,for a total of366s of test audio.The MIREX 2005melody evaluation created a new test set consisting of25excerpts ranging in length from10-40s,giving536s total.Both sets include excerpts where the domi-nant melody is played on a variety of musical instruments including the human voice; the MIREX set has more of a bias to pop music.In this paper,all of our experiments are conducted using the ADC2004data as a development set,since the MIREX set is reserved for annual evaluations3Acoustic FeaturesOur acoustic representation is based on the ubiquitous and well-known spectrogram, which converts a sound waveform into a distribution of energy over time and frequency, very often displayed as a pseudocolor or grayscale image like the middle pane infigure 1;the base features for each time-frame can be considered as vertical slices through such an image.Specifically,the original music recordings(melody plus accompani-ment)are combined into a single(mono)channel and downsampled to8kHz.We apply the short-time Fourier transform(STFT),using N=1024point transforms(i.e. 128ms),an N-point Hanning window,and a944point overlap of adjacent windows (for a10ms hop between successive frames).Only the coefficients corresponding to frequencies below2kHz(i.e.thefirst255bins)were used in the feature vector.We compared different feature preprocessing schemes by measuring their influ-ence on a baseline classifier.Our baseline pitch classifier is an all-versus-all(A V A) algorithm for multiclass classification using SVMs trained by Sequential Minimal Op-timization[Platt,1998],as implemented in the Weka toolkit[Witten and Frank,2000]. In this scheme,a majority vote is taken from the output of(N2−N)/2discriminant functions,comparing every possible pair of classes.For computational reasons,we were restricted to a linear kernel.Each audio frame is represented by a256-element input vector,with N=60classes corresponding tofive-octaves of semitones from G2 to F#7.In order to classify the dominant melodic pitch for each frame,we assume the melody note at a given instant to be solely dependent on the normalized frequency data below2kHz.For these results,we further assume each frame to be independent of all other frames.More details and experiments concerning the classifier will be presented in section4.Separate classifiers were trained using six different feature normalizations.Of these,three use the STFT,and three are based on(pseudo)autocorrelation.In thefirst case,we simply used the magnitude of the STFT normalized such that the maximum energy frame in each song had a value equal to one.For the second case,the magni-tudes of the bins are normalized by subtracting the mean and dividing by the standard deviation calculated in a71-point sliding frequency window;there is no normalization along time.The goal is to remove some of the influence due to different instrument timbres and contexts in training and test data.The third normalization scheme applied cube-root compression to the STFT magnitude,to make larger spectral magnitudes ap-pear more similar;cube-root compression is commonly used as an approximation to6Table1:Effect of normalization:Frame accuracy percentages on the ADC2004held-out test set for each of the normalization schemes considered,trained on either multi-track audio alone,MIDI syntheses alone,or both data sets combined.(Training sets are roughly balanced in size,so results are not directly comparable to others in the paper.)Training dataNormalization Multi-track MIDI BothSTFT56.450.562.571-pt norm54.246.162.7Cube root53.351.262.4Autocorr55.845.262.4Cepstrum49.345.254.6LiftCeps55.845.362.3the loudness sensitivity of the ear.A fourth feature took the inverse Fourier transform(IFT)of the magnitude of the STFT,i.e.the autocorrelation of the original windowed waveform.Taking the IFT of the log-STFT-magnitude gives the cepstrum,which comprised ourfifth feature type. Because overall gain and broad spectral shape are separated into thefirst few cepstral bins,whereas periodicity appears at higher indexes,this feature also performs a kind of timbral normalization.We also tried normalizing these autocorrelation-based features by liftering(scaling the higher-order cepstra by an exponential weight).Scaling of individual feature dimensions can make a difference to SVM classifiers,depending on the kernel function used.Table1compares the accuracy,on a held-out test set,of classifiers trained on each of the different normalization schemes.Here we show separate results for the classi-fiers trained on multi-track audio alone,MIDI syntheses alone,or both data sources combined.The frame accuracy results are for the ADC2004melody evaluation set and correspond to melodic pitch transcription to the nearest semitone.The most obvious result in table1is that all the features,with the exception of Cepstrum,perform much the same,with a slight edge for the across-frequency local normalization.This is perhaps not surprising since all features contain largely equiv-alent information,but it also raises the question as to how effective our normalization (and hence the system generalization)has been.It may be that a better normalization scheme remains to be discovered.Looking across the columns in the table,we see that the more realistic multi-track data forms a better training set than the MIDI syntheses,which have much lower acous-tic similarity to most of the evaluation ing both,and hence a more diverse training set,always gives a significant accuracy boost–up to9%absolute improve-ment,seen for the best-performing71-point normalized features.(We will discuss significance levels for these results in the next section.)Table2shows the impact of including the replicas of the training set transposed through resampling over±6semitones.Resampling achieves a substantial improve-7Table2:Impact of resampling the training data:Frame accuracy percentages on the ADC2004set for systems trained on the entire training set,either without any resam-pling transposition,or including transpositions out to±6semitones(500frames per transposed excerpt,17excerpts,1or13transpositions).Training set#Training frames Frame acc%No resampling8,50060.2With resampling110,50067.7ment of7.5%absolute in frame accuracy,underlining the value of broadening the range of data seen for each individual note.4Pitch ClassificationIn the previous section,we showed that classification accuracy seems to depend more strongly on training data diversity than on feature normalization.It may be that the SVM classifier we used is better able to generalize than our explicit feature normaliza-tion.In this section,we examine the effects of different classifier types on classification accuracy,as well as looking at the influence of the total amount of training data used.4.1N-way All-Versus-All SVM ClassificationOur baseline classifier is the A V A-SVM described in section3above.Given the large amount of training data we were using(over105frames),we chose a linear kernel, which requires training time on the order of the number of feature dimensions cubed for each of the O(N2)discriminant functions.More complex kernels(such as Radial Basis Functions,which require training time on the order of the number of instances cubed)were computationally infeasible for our large training set.Recall that labels are assigned independently to each time frame at this stage of processing.Ourfirst classification experiment was to determine the number of training in-stances to include from each audio excerpt.The number of training instances selected from each song was varied using both incremental sampling(taking a limited num-ber of frames from the beginning of each excerpt)and random sampling(picking the frames from anywhere in the excerpt),as displayed infigure2.Randomly sampling feature vectors to train on approaches an asymptote much more rapidly than adding the data in chronological order.Random sampling also appears to exhibit symptoms of overtraining.The observation that random sampling achieves peak accuracy with only about 400samples per excerpt(out of a total of around3000for a30s excerpt with10ms hops)can be explained by both signal processing and musicological considerations. Firstly,adjacent analysis frames are highly overlapped,sharing118ms out of a128ms window,and thus their feature values will be very highly correlated(10ms is an un-necessarilyfine time resolution to generate training frames,but is the standard used in83035404550556065Training samples per excerpt (17 excerpts total)C l a s s i f i c a t i o n a c c u r a c y / %Figure 2:Variation of classification accuracy with number of training frames per ex-cerpt.Incremental sampling takes frames from the beginning of the excerpt;random sampling takes them from anywhere.Training set does not include resampled (trans-posed)data.the evaluation).From a musicological point of view,musical notes typically maintain approximately constant spectral structure over hundreds of milliseconds;a note should maintain a steady pitch for some significant fraction of a beat to be perceived as well-tuned.If we assume there are on average 2notes per second (i.e.around 120bpm)in our pop-based training data,then we expect to see approximately 60melodic note events per 30s excerpt.Each note may contribute a few usefully different frames to tuning variation such as vibrato and variations in accompaniment.Thus we expect many clusters of largely redundant frames in our training data,and random sampling down to 10%(or closer to one frame every 100ms)seems reasonable.This also gives us a perspective on how to judge the significance of differences in these results.The ADC 2004test set consists of 366s,or 36,600frames using the standard 10ms hop.A simple binomial significance test can compare classifiers by es-timating the likelihood that random sets of independent trials could give the observed differences in empirical error rates from an equal underlying probability of error.Since the standard error of such an observation falls as 1/√N for N trials,the significance interval depends directly on the number of trials.However,the arguments and ob-servations above show that the 10ms frames are anything but independent;to obtain something closer to independent trials,we should test on frames no less than 100ms apart,and 200ms sampling (5frames per second)would be a safer choice.This corre-sponds to only 1,830independent trials in the test set;a one-sided binomial significance test suggests that differences in frame accuracies on this test of less than 2.5%are not statistically significant at the accuracies reported in this paper.A second experiment examined the incremental gain from adding novel training excerpts.Figure 3shows how classification accuracy increased as increasing numbers of excerpts,from 1to 16,were used for training.In this case,adding an excerpt con-sisted of adding 500randomly-selected frames from each of the 13resampled transpo-9Number of training frames (500 x 13 = 6500 per excerpt)Training Data + Resampled AudioC l a s s i f i c a t i o n a c c u r a c y / %Figure 3:Variation of classification accuracy with the total number of excerpts in-cluded,compared to sampling the same total number of frames from all excerpts pooled.This data set includes 13resampled versions of each excerpt,with 500frames randomly sampled from each transposition.sitions described in section 2,or 6,500frames per excerpt.Thus,the largest classifier is trained on 104k frames,compared to around 15k frames for the largest classifier in figure 2.The solid curve shows the result of training the same number of frames ran-domly drawn from the pool of the entire training set;again,we notice that the system appears to reach an asymptotes by 20k total frames,or fewer than 100frames per trans-posed excerpt.We believe,however,that the level of this asymptote is determined by the total number of excerpts;if we had more novel training data to include,we believe that the “per excerpt”trace will continue to climb upwards.We return to this point in the discussion.In figure 4,the pitched frame transcription success rates are displayed for the SVM classifier trained using the resampled audio.An important weakness of the classifier-based approach is that any classifier will perform unpredictably on test data that does not resemble the training data.While model-based approaches have no problem in principle with transcribing rare,extreme pitches as long as they conform to the explicit model,by deliberately ignoring our expert knowledge of the relationship between spec-tra and notes our system is unable to generalize from the notes it has seen to different pitches.For example,the highest pitch values for the female opera samples in the ADC 2004test set exceed the maximum pitch in all our training data.In addition,the ADC 2004set contains stylistic genre differences (such as opera)that do not match our pop music corpora.That said,many of the pitch errors turn out to be the right pitch chroma class but at the wrong octave (e.g.F6instead of F7).When scoring is done on chroma class alone –ing only the 12classes A,A#,...G,and ignoring the octave num-bers –the overall frame accuracy on the ADC 2004test set improves from 67.7%to10daisy jazz midi opera popf r a m e a c c u r a c y % - c h r o m a f r a m e a c c u r a c y % - r a w Figure 4:Variation in voiced transcription frame accuracy across the 20excerpts of the ADC 2004evaluation set (4examples from each of 5genre categories,as shown).Solid line shows the classification-based transcriber;dashed line shows the results of the best-performing system from the 2004evaluation.Top pane is raw pitch accuracy;bottom pane folds all results to a single octave of 12chroma bins,to ignore octave errors.72.7%.4.2Multiple One-Versus-All SVM ClassifiersIn addition to the N-way melody classification,we trained 52binary,one-versus-all (OV A)SVM classifiers representing each of the notes present in the resampled training set.We took the distance-to-classifier-boundary hyperplane margins as a proxy for a log-posterior probability for each of these classes;pseudo-posteriors (up to an arbitrary scaling power)were obtained from the distances by fitting a logistic model.Transcrip-tion is achieved by choosing the most probable class at each time frame.While OV A approaches are seen as less sophisticated,Rifkin and Klautau [2004]present evidence that they can match the performance of more complex multiway classification schemes.Figure 5shows an example ‘posteriorgram’(time-versus-class image showing the pos-teriors of each class at each time step)for a pop excerpt,with the ground truth labels overlaid.Since the number of classifiers required for this task is O (N )(unlike the O (N 2)classifiers required for the A V A approach)it becomes computationally feasible to ex-periment with additional feature kernels.Table 3displays the best result classification rates for each of the SVM classifiers.Both OV A classifiers perform marginally better than the pairwise classifier,with the slight edge going to the OV A SVM using an RBF kernel.11。