├── README.md ├── US20160189730A1_speech_separation_patent.pdf ├── YongXU_Taslp_2015.pdf ├── Yong_SPL_SEDNN_2013.pdf ├── _config.yml ├── audio.mp3 ├── audio.wav ├── config ├── readme.txt └── timit_115NT_7SNRs_each190utts_noisy_lspANDmfcc93b40_be_random_linux_global_mv.mat ├── help ├── HCopy.exe ├── LogSpec2Wav_lenMulti20.exe ├── LogSpec2raw_16bit_withoutXF.exe ├── Raw2Wav.exe ├── WAV2RAW.exe ├── Wav2LogSpec.exe ├── config_mel40.mfcc ├── frame_expand.m ├── msvcr100.dll ├── msvcr100d.dll ├── readhtk_new.m ├── readme.txt ├── step1_DNNenh_for16kHz.m ├── step1_wav2raw.pl ├── step2_get_scp_raw.pl ├── step3_get_lsp.pl ├── step4_get_scp_lsp.pl ├── tmp │ ├── logspec2wav │ │ └── readme.txt │ └── wav2logspec │ │ └── readme.txt └── writeHTK_new.m ├── how to run.txt ├── interspeech2015_yongxu.pdf ├── run.bat ├── scp ├── readme.txt └── yqzu_lsp_16k_noisy.scp ├── step1_DNNenh_for16kHz.m ├── tmp ├── DNN_enh.raw ├── DNN_enh.wav ├── info.txt ├── noisy.mfcc ├── noisy.wav ├── out.htk ├── readme.txt ├── yqzu_lsp_16k_noisy.scp └── yqzu_raw_16k_noisy.scp ├── video.mp4 └── wav_lsp ├── readme.txt ├── timit_noisy_SNR5.lsp ├── timit_noisy_SNR5.raw ├── timit_noisy_SNR5.wav └── timit_noisy_SNR5_DNNenh.wav /README.md: -------------------------------------------------------------------------------- 1 | 2 | This code is only for research. If you want to use it for commercial reason, please contact me: yong.xu.ustc@gmail.com 3 | 4 | # DNN-Speech-enhancement-demo-tool 5 | Universal Deep neural network based speech enhancement demo and tools, well pre-trained DNN model 6 | 7 | the model/weights can be downloaded here: https://drive.google.com/file/d/1Wg14r0m41kWv2ja-stBkq2uZgjOP5yME/view?usp=sharing or https://drive.google.com/file/d/0B5r5bvRpQ5DRR1lIV1hpZ0RLQ0E/view?usp=sharing 8 | also in the config/* , or you can download from (@ Baidu Yun) http://pan.baidu.com/s/1eRJGrx4 9 | 10 | GPU code for Deep neural network (DNN) based speech enhancement can be found here: https://github.com/yongxuUSTC/DNN-for-speech-enhancement 11 | 12 | What kinds of noisy speech can the DNN-enh tool enhance ? 13 | 14 | It can enhance any kinds of noisy speech, even the real-world noisy speech. one real-world noisy speech enh demo for the movie : http://staff.ustc.edu.cn/~jundu/The%20team/yongxu/demo/IS15.html 15 | 16 | The model is trained only on TIMIT data, so it can get the best performance on the TIMIT test set. 17 | 18 | The model can get the best performance on English dataset because TIMIT is US-English. But this tool still can be used to enhance the noisy speech in other languages, like Chinese. 19 | 20 | You can use multi-language data to retrain this model to get a general DNN-enh tool. 21 | 22 | What else can this code use for? 23 | 24 | It is designed for any regression tasks, like speech enhancement, ideal binary/ratio mask (IBM/IRM) estimation, audio/music tagging, acoustic event detection, etc. 25 | Please cite the following papers if you use this code: 26 | 27 | [1]A Regression Approach to Speech Enhancement Based on Deep Neural Networks.YongXu,JunDu,Li-Rong Dai and Chin-Hui Lee, IEEE/ACM Transactions on Audio,Speech, and Language Processing,P.7-19,Vol.23,No.1, 2015 (2018 IEEE SPS Best paper award, citations > 350) 28 | 29 | [2]An Experimental Study on Speech Enhancement Based on Deep Neural Networks.YongXu, JunDu, Li-Rong Dai and Chin-Hui Lee,IEEE signal processing letters, p. 65-68,vol.21,no. 1,January 2014 30 | 31 | [3] Multi-Objective Learning and Mask-Based Post-Processing for Deep Neural Network Based Speech Enhancement, Yong Xu, Jun Du, Zhen Huang, Li-Rong Dai, Chin-Hui Lee, Interspeech2015 32 | 33 | Some DNN based speech enhancemen demos: 34 | 35 | http://staff.ustc.edu.cn/~jundu/The%20team/yongxu/demo/SE_DNN_taslp.html http://staff.ustc.edu.cn/~jundu/The%20team/yongxu/demo/IS15.html 36 | 37 | [](https://youtu.be/Z4quxzXR5yw) 38 | 39 | 40 | 41 | 42 | -------------------------------------------------------------------------------- /US20160189730A1_speech_separation_patent.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/US20160189730A1_speech_separation_patent.pdf -------------------------------------------------------------------------------- /YongXU_Taslp_2015.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/YongXU_Taslp_2015.pdf -------------------------------------------------------------------------------- /Yong_SPL_SEDNN_2013.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/Yong_SPL_SEDNN_2013.pdf -------------------------------------------------------------------------------- /_config.yml: -------------------------------------------------------------------------------- 1 | theme: jekyll-theme-slate -------------------------------------------------------------------------------- /audio.mp3: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/audio.mp3 -------------------------------------------------------------------------------- /audio.wav: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/audio.wav -------------------------------------------------------------------------------- /config/readme.txt: -------------------------------------------------------------------------------- 1 | the model/weights can be downloaded here: https://drive.google.com/file/d/0B5r5bvRpQ5DRR1lIV1hpZ0RLQ0E/view?usp=sharing 2 | also in the config/* , or you can download from (@ Baidu Yun) http://pan.baidu.com/s/1eRJGrx4 3 | -------------------------------------------------------------------------------- /config/timit_115NT_7SNRs_each190utts_noisy_lspANDmfcc93b40_be_random_linux_global_mv.mat: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/config/timit_115NT_7SNRs_each190utts_noisy_lspANDmfcc93b40_be_random_linux_global_mv.mat -------------------------------------------------------------------------------- /help/HCopy.exe: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/help/HCopy.exe -------------------------------------------------------------------------------- /help/LogSpec2Wav_lenMulti20.exe: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/help/LogSpec2Wav_lenMulti20.exe -------------------------------------------------------------------------------- /help/LogSpec2raw_16bit_withoutXF.exe: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/help/LogSpec2raw_16bit_withoutXF.exe -------------------------------------------------------------------------------- /help/Raw2Wav.exe: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/help/Raw2Wav.exe -------------------------------------------------------------------------------- /help/WAV2RAW.exe: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/help/WAV2RAW.exe -------------------------------------------------------------------------------- /help/Wav2LogSpec.exe: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/help/Wav2LogSpec.exe -------------------------------------------------------------------------------- /help/config_mel40.mfcc: -------------------------------------------------------------------------------- 1 | SOURCEKIND = WAVEFORM 2 | SOURCEFORMAT = WAVE 3 | SOURCERATE = 625 4 | TARGETKIND = MFCC_0_D_A_Z 5 | TARGETRATE = 160000.0 6 | WINDOWSIZE = 320000.0 7 | USEHAMMING = T 8 | PREEMCOEF = 0.97 9 | NUMCHANS = 40 10 | #CEPLIFTER = 22 11 | NUMCEPS = 30 12 | -------------------------------------------------------------------------------- /help/frame_expand.m: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/help/frame_expand.m -------------------------------------------------------------------------------- /help/msvcr100.dll: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/help/msvcr100.dll -------------------------------------------------------------------------------- /help/msvcr100d.dll: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/help/msvcr100d.dll -------------------------------------------------------------------------------- /help/readhtk_new.m: -------------------------------------------------------------------------------- 1 | %% ************************************************************************ 2 | % readHTK - just incase you ever want to go backwards 3 | %************************************************************************** 4 | function [htkdata,nframes,sampPeriod,sampSize,paramKind] = readhtk_new(filename,byte_order) 5 | 6 | if nargin<2 7 | byte_order = 'be'; 8 | end 9 | 10 | fid = fopen(filename,'r',sprintf('ieee-%s',byte_order)); 11 | 12 | nframes = fread(fid,1,'int32'); 13 | sampPeriod = fread(fid,1,'int32'); 14 | sampSize = fread(fid,1,'int16'); 15 | paramKind = fread(fid,1,'int16'); 16 | 17 | % read the data 18 | 19 | htkdata = fread(fid,nframes*(sampSize/4),'float32'); 20 | htkdata = reshape(htkdata,sampSize/4,nframes); 21 | fclose(fid); 22 | end % ------ OF READHTK -------------------------------------------------------------------------------- /help/readme.txt: -------------------------------------------------------------------------------- 1 | some basic tools 2 | -------------------------------------------------------------------------------- /help/step1_DNNenh_for16kHz.m: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/help/step1_DNNenh_for16kHz.m -------------------------------------------------------------------------------- /help/step1_wav2raw.pl: -------------------------------------------------------------------------------- 1 | use warnings; 2 | 3 | ##noisy 4 | $inpath="wav_lsp"; 5 | $outpath="wav_lsp"; 6 | 7 | if (!(-e $outpath)) {system("mkdir $outpath");} 8 | 9 | #$inpath="D\:\\user\\jundu\\xuyong\\yanping_10h\\raw_8k"; 10 | #$outpath="D\:\\user\\jundu\\xuyong\\yanping_10h\\lsp_8k"; 11 | 12 | 13 | #open(OUT1,">scp\\yanping_all_lsp_babble_8k_N2SNR10.scp"); 14 | #open(OUT1,">scp\\yanping_all_lsp_8k_clean.scp"); 15 | opendir(DIR,$inpath); 16 | foreach $f (readdir DIR) 17 | { 18 | if($f=~m/^\./){next;} 19 | $outf=$f; 20 | $outf=~s/\.wav/\.raw/; 21 | 22 | system("help\\WAV2RAW.exe $inpath\\$f $outpath\\$outf"); 23 | # print OUT1 "$outpath\\$outf\n"; 24 | } 25 | 26 | ###clean 27 | #$inpath="D:\\user\\jundu\\xuyong\\cmu_one_speaker\\CMU_ARCTIC\\cmu_us_slt_arctic-0.95-release\\cmu_us_slt_arctic\\raw_8k"; 28 | #$outpath="D:\\user\\jundu\\xuyong\\cmu_one_speaker\\CMU_ARCTIC\\cmu_us_slt_arctic-0.95-release\\cmu_us_slt_arctic\\lsp_8k"; 29 | # 30 | #open(OUT1,">cmu_arctic_female_lsp_clean_8k.scp"); 31 | #opendir(DIR,$inpath); 32 | #foreach $f (readdir DIR) 33 | #{ 34 | # if($f=~m/^\./){next;} 35 | # $outf=$f; 36 | # $outf=~s/\.raw/\.lsp/; 37 | # 38 | # system("D\:\\user\\jundu\\AE_SE\\Bin\\Wav2LogSpec.exe -F RAW -fs 8 $inpath\\$f $outpath\\$outf"); 39 | # print OUT1 "$outpath\\$outf\n"; 40 | #} -------------------------------------------------------------------------------- /help/step2_get_scp_raw.pl: -------------------------------------------------------------------------------- 1 | use warnings; 2 | 3 | #my ($path, $oscp) = @ARGV; 4 | $path="wav_lsp"; 5 | $oscp="tmp\\yqzu_raw_16k_noisy.scp"; 6 | 7 | open(OUT1,">$oscp"); 8 | opendir(DIR,$path); 9 | foreach $f (readdir DIR) 10 | { 11 | if($f=~m/\.raw/) 12 | { 13 | print OUT1 "$path\\$f\n"; 14 | } 15 | } 16 | -------------------------------------------------------------------------------- /help/step3_get_lsp.pl: -------------------------------------------------------------------------------- 1 | use warnings; 2 | 3 | $scp="tmp\\yqzu_raw_16k_noisy.scp"; 4 | open(IN1,$scp); 5 | while() 6 | { 7 | chomp; 8 | $line=$_; 9 | if($line=~m/\.raw/) 10 | { 11 | $inline=$line; 12 | $outline=$line; 13 | $outline=~s/\.raw/\.lsp/; 14 | system("help\\Wav2LogSpec.exe -F RAW -fs 16 $inline $outline"); 15 | } 16 | 17 | #die; 18 | } 19 | -------------------------------------------------------------------------------- /help/step4_get_scp_lsp.pl: -------------------------------------------------------------------------------- 1 | use warnings; 2 | 3 | #my ($path, $oscp) = @ARGV; 4 | $path="wav_lsp"; 5 | $oscp="tmp\\yqzu_lsp_16k_noisy.scp"; 6 | 7 | open(OUT1,">$oscp"); 8 | opendir(DIR,$path); 9 | foreach $f (readdir DIR) 10 | { 11 | if($f=~m/\.lsp/) 12 | { 13 | print OUT1 "$path\\$f\n"; 14 | } 15 | } 16 | system("copy tmp\\yqzu_lsp_16k_noisy.scp scp\\yqzu_lsp_16k_noisy.scp"); 17 | -------------------------------------------------------------------------------- /help/tmp/logspec2wav/readme.txt: -------------------------------------------------------------------------------- 1 | as title 2 | -------------------------------------------------------------------------------- /help/tmp/wav2logspec/readme.txt: -------------------------------------------------------------------------------- 1 | as title 2 | -------------------------------------------------------------------------------- /help/writeHTK_new.m: -------------------------------------------------------------------------------- 1 | %% ************************************************************************ 2 | % writeHTK - convert data to htk format -> by anonymous Sue (2001) 3 | %************************************************************************** 4 | function retcode = writeHTK_new(filename, htkdata, nFrames, sampPeriod, SampSize, ParamKind, byte_order) 5 | % Write an HTK format file. 6 | % 7 | % Input parameters: 8 | % filename HTK data file 9 | % htkdata HTK data read: an m x n matrix with 10 | % m = no. of channels 11 | % n = no. of frames 12 | % The following are from the HTK header (see HTK manual): 13 | % nFrames no. of frames (samples) 14 | % sampPeriod sample period (in 100 ns units?) 15 | % SampSize sample size 16 | % ParamKind parameter kind code 17 | % 18 | % byteorder 'be' for big-endian (typical for Unix) (default) 19 | % 'le' for little-endian (typical for MSWindows) 20 | % 21 | % Output parameters: 22 | % retcode 0 if successful 23 | 24 | % Written by Sue 17/12/01 25 | 26 | retcode=-1; % initialise in case of error 27 | if nargin < 6 28 | fprintf('Usage: %s(filename, htkdata, nFrames, sampPeriod, SampSize, ParamKind [, byte_order])', mfilename); 29 | end; 30 | 31 | % Default to big-endian (HTK format) 32 | if nargin < 7 33 | byte_order = 'be'; 34 | end; 35 | 36 | fid = fopen (filename, 'w', sprintf('ieee-%s', byte_order)); 37 | if fid < 1 38 | fprintf('%s: can''t open output file %s\n', mfilename, filename); 39 | return 40 | end 41 | 42 | % Write header 43 | fwrite (fid, nFrames, 'int32'); %nSamples in HTK 44 | fwrite (fid, sampPeriod, 'int32'); 45 | fwrite (fid, SampSize, 'int16'); 46 | fwrite (fid, ParamKind, 'int16'); 47 | 48 | % Write actual data 49 | fwrite(fid, htkdata', 'float32'); 50 | 51 | fclose(fid); 52 | 53 | retcode=0; 54 | end% ------ OF WRITEHTK 55 | -------------------------------------------------------------------------------- /how to run.txt: -------------------------------------------------------------------------------- 1 | 2 | DNN based speech enhancement tool (developed by Dr. yong xu) 3 | 4 | Note: Matlab should be installed in your PC first. Perl should be installed in your PC. 5 | This zip contains the best DNN model trained by us. 6 | This exe can be used to enhance any noisy speech, even from real world noisy speech! 7 | 8 | Enhancing steps: 9 | 10 | 1 step0: clear all of the files in "wav_lsp", and then copy all of your noisy waveforms into "wav_lsp" 11 | 12 | 2 step1: in Matlab, JUST RUN "step1_DNNenh_for16kHz.m" without any parameters to be tuned and then all of the noisy waveforms will be obtained in "wav_lsp" 13 | (There is a noisy 16kHz waveform in wav_lsp/timit_noisy_SNR5.wav as an example) 14 | 15 | or you can just click: step1_DNNenh_for16kHz.exe, everything will be done for you! 16 | 17 | 18 | 19 | 20 | all is done!!! 21 | 22 | The related enhanced demos: 23 | http://home.ustc.edu.cn/~xuyong62/demo/IS15.html 24 | http://home.ustc.edu.cn/~xuyong62/demo/SE_DNN_taslp.html 25 | 26 | If you have any questions, please feel free to contact Dr. Yong XU (yong.xu.ustc@gmail.com) 27 | 28 | Please cite: 29 | [1] Multi-objective learning and Mask-based Post-processing for Deep Neural Network based Speech Enhancement. 30 | Yong Xu, Jun Du, Zhen Huang, Li-Rong Dai, Chin-Hui Lee, accepted, Interspeech2015, Dresden, Germany 31 | [2] A Regression Approach to Speech Enhancement Based on Deep Neural Networks. 32 | Yong Xu, Jun Du, Li-Rong Dai and Chin-Hui Lee, accepted to IEEE/ACM Transactions on Audio, Speech, and Language Processing, P. 7-19, Vol. 33 | 23, No. 1, IEEE/ACM trans. on audio, speech, language processing, 2015 34 | [3] An Experimental Study on Speech Enhancement Based on Deep Neural Networks. 35 | Yong Xu, Jun Du, Li-Rong Dai and Chin-Hui Lee, IEEE signal processing letters, p. 65-68, vol. 21, no. 1, January 2014 36 | [4] SPEECH SEPARATION METHOD AND SYSTEM, US patent, US 20160189730A1 37 | 38 | Note: if you want to use it for commercial product, please contact Dr. Yong XU (yong.xu.ustc@gmail.com) -------------------------------------------------------------------------------- /interspeech2015_yongxu.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/interspeech2015_yongxu.pdf -------------------------------------------------------------------------------- /run.bat: -------------------------------------------------------------------------------- 1 | step1_DNNenh_for16kHz.exe -------------------------------------------------------------------------------- /scp/readme.txt: -------------------------------------------------------------------------------- 1 | scp 2 | -------------------------------------------------------------------------------- /scp/yqzu_lsp_16k_noisy.scp: -------------------------------------------------------------------------------- 1 | wav_lsp\timit_noisy_SNR5.lsp 2 | -------------------------------------------------------------------------------- /step1_DNNenh_for16kHz.m: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/step1_DNNenh_for16kHz.m -------------------------------------------------------------------------------- /tmp/DNN_enh.raw: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/tmp/DNN_enh.raw -------------------------------------------------------------------------------- /tmp/DNN_enh.wav: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/tmp/DNN_enh.wav -------------------------------------------------------------------------------- /tmp/info.txt: -------------------------------------------------------------------------------- 1 | Segmental SNR: 2 | 3.392618 3 | Log-Spectral Distortion: 4 | 10.548547 5 | -------------------------------------------------------------------------------- /tmp/noisy.mfcc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/tmp/noisy.mfcc -------------------------------------------------------------------------------- /tmp/noisy.wav: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/tmp/noisy.wav -------------------------------------------------------------------------------- /tmp/out.htk: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/tmp/out.htk -------------------------------------------------------------------------------- /tmp/readme.txt: -------------------------------------------------------------------------------- 1 | tmp files 2 | -------------------------------------------------------------------------------- /tmp/yqzu_lsp_16k_noisy.scp: -------------------------------------------------------------------------------- 1 | wav_lsp\timit_noisy_SNR5.lsp 2 | -------------------------------------------------------------------------------- /tmp/yqzu_raw_16k_noisy.scp: -------------------------------------------------------------------------------- 1 | wav_lsp\timit_noisy_SNR5.raw 2 | -------------------------------------------------------------------------------- /video.mp4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/video.mp4 -------------------------------------------------------------------------------- /wav_lsp/readme.txt: -------------------------------------------------------------------------------- 1 | wav and fea 2 | -------------------------------------------------------------------------------- /wav_lsp/timit_noisy_SNR5.lsp: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/wav_lsp/timit_noisy_SNR5.lsp -------------------------------------------------------------------------------- /wav_lsp/timit_noisy_SNR5.raw: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/wav_lsp/timit_noisy_SNR5.raw -------------------------------------------------------------------------------- /wav_lsp/timit_noisy_SNR5.wav: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/wav_lsp/timit_noisy_SNR5.wav -------------------------------------------------------------------------------- /wav_lsp/timit_noisy_SNR5_DNNenh.wav: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/yongxuUSTC/DNN-Speech-enhancement-demo-tool/ad053991d768557a055205fe8539f44869fa6f7a/wav_lsp/timit_noisy_SNR5_DNNenh.wav --------------------------------------------------------------------------------