Meyda特征提取数据在命令行和nodejs中有所不同

Question

我尝试使用Meyda提取音频文件的特性。但是，当使用命令行meyda和通过javascript文件提取时，特征提取值是不同的

1. 命令行

   meyda 1600153162.63571.wav mfcc

结果

0.057628476337413304,0.025103534166687494、-0.0033402018588145435、-0.00050914619032203364、-0.005969043107840086、-0.0094239628697326、-0.007781823184730455、-0.002403016732702066、-0.0033368248853514206、-0.001272591871309979,0.00003587458387239886,0.002937354749062950.0003162315327070361.4306079393227265、1.3713938566968262、0.2306487569743767、1.0515807610571597、0.8414768385049735、.6060415036630535、0.37326507523088365、0.1654296077992904、0.015392187854876595、0.16525095936577213、0.27571446879975764、0.340564139593766、-0.3650538438926707039.282664122642018、6.880236029290547、.3870794551074606、1.7170683624244123、-0.9657914336587826、-318544774959499、4.433789386396694、3.7101490665746217、2.816738558429225、2.2736417577176273、-1.636438686466657、-0.537746809738949、-0.5116460308902115225.65632094501052,22.93621824737729,16.08611763344924,7.490656903754756，-0.358888007314062，-5.64156791337735，-7.229285832273481，-5.63856022395436，2.1681330598896817,1.5575050036371694,4.296381195632149,5.45750656573637,5.05996621856976823.469876801827922,20.984500104582974,15.80549272827184,10.064116650359527,3.957435073588796,1.75506262168123,5.904087102697238,8.29320895086182,9.037630914839253,8.230609293960022,6.6041597934477832,4.8505887091085365-3.23906727518617529.1420600661076625.50517483703385,19.279159148659456,13.806566615476791,7.715092034538,2.2364700767576764，0.8129542016331001，2.16772108652068，2.6257181917727093，2.1347711722007845，1.0689231846996041，0.6059541306434987，-0.9955565080584865112.251538716256618,9.479093353756676,5.2726325506769,2.7893618666871762,0.10288164291841367，3.121859159330943，4.446188894370852，3.5959072906303837，2.5699522942015007，1.8300687753240275，0.7140618177046422,0.3644613187160053，0.835318506508760418.42772721964866,15.9904592445750096,11.9931328734504,9.088259023158539，.1723551883886699,2.6632784721745213，-0.683186290158647，1.948275236251193，2.793271841187546，3.305172382572076，3.6707865100252994，3.6389999985131219，-2.86326382282771515.915610973257571,13.737073178518493,9.801934620786648,5.779403187264773,1.358952752885622，-2.50841159816647，-5.187471901991046，-7.033404192058985，-7.737140523422115，-7.437282553994893，-6.058991955183858，-4.36024675428363，-2.68062980539606615.75090683856979,13.4970898998444463,9.489154494388005,5.57883842168272,1.4567586737610156，2.141429668615224，-4380728646793593，5.364310654901947，5.049848052960211，4.214979808517224，3.9745754014376224，4.076571890925421，-4.02071430360983423.361729244701564,20.83103985201436,16.592090051349242,12.885538444506535,8.373777248668826,2.9768227085972145，-1399644048138661，-4.0944050595665753，5.574775136428788，-5.787457864519816，-4.693018758118676，-3100993535744204，-1.562554637739947512.793707716744393,10.197116076581045,5.819041457288568,2.4679725003955064，-0.3304198215142393，-2.607817847380143，-4.086110143731154，-4.79218992884254，-4.3744659496245，-3118504640745296，-1.666020102081847，-1.51407789377241373，-1.51306470109503133.2732387410942465,2.2453286293698977,0.9749483649128511,0.472715919767034,0.11249067642678866，-0.1927487640318992，-0.242623087509772，0.3645480814147444，0.5060378572115204，-0.793996411313405，-0.5934442588699886，-0.68076105511004，-0.2967506177492105

2. 通过javscript文件

   var express = require('express');
   var app = express();
   const fs = require('fs');
   var Meyda = require('meyda');
   var load = require('audio-loader')
   let filename  = "1600153162.63571.wav"
   load(filename).then(function (buffer) {
   const channelData = buffer.getChannelData(0) 
   const PaddingLength = (Math.pow(2,Math.round(Math.log2(channelData.length)+1)) - channelData.length)
   let halfPaddingLength = parseInt(PaddingLength/2)
   const pad1 = new Array(halfPaddingLength).fill(0);
   const pad2 = new Array(PaddingLength - halfPaddingLength).fill(0);
   let finalBbuffer = [...pad1,...channelData,...pad2]
   console.log(finalBbuffer.length)
   let mfccData =  Meyda.extract('mfcc', finalBbuffer)
   console.log("mfccData : ",mfccData);
   });
   

结果：

[249.11783051490784，-90.61751411189829，-12.253094024524968，19.88245460444982，-11.661965456271869，-14.795375019626466，7.19298966815922，8.362884489124907，3.9941283332736557，-3.21583888656478287，-2.0395393071161063，-0.488491767514482837，-3.619963273626695]

为什么会有这种差异。如何使它们相同？

Answer 1

它们不同，因为您在代码中填充缓冲区，而命令行工具将文件拆分为具有默认缓冲区大小的多个缓冲区，并为每个段返回MFCC。

为了使它们相同，您可以复制CLI在代码中执行的缓冲区分块，而不是对整个缓冲区进行零填充，或者您可以制作一个对输入进行零填充的CLI。如果您选择后一种方式，我们当然有兴趣将该功能添加到捆绑的CLI中，因此请随时在我们的问题跟踪器上打开一个问题，讨论您是否愿意这样做并做出贡献。

感谢您使用Meyda！