swshon / Voxceleb Ivector
Licence: mit
Voxceleb1 i-vector based speaker recognition system
Projects that are alternatives of or similar to Voxceleb Ivector
kaldi-allignerscripts to align a given wave to its transcription using trained models by Kaldi
Stars: ✭ 24 (-33.33%)
Mutual labels: kaldi
React Transcript EditorA React component to make correcting automated transcriptions of audio and video easier and faster. By BBC News Labs. - Work in progress
Stars: ✭ 285 (+691.67%)
Mutual labels: kaldi
EesenThe official repository of the Eesen project
Stars: ✭ 738 (+1950%)
Mutual labels: kaldi
speech-to-textmixlingual speech recognition system; hybrid (GMM+NNet) model; Kaldi + Keras
Stars: ✭ 61 (+69.44%)
Mutual labels: kaldi
Vosk Android DemoOffline speech recognition for Android with Vosk library.
Stars: ✭ 271 (+652.78%)
Mutual labels: kaldi
EspnetEnd-to-End Speech Processing Toolkit
Stars: ✭ 4,533 (+12491.67%)
Mutual labels: kaldi
opensnipsOpen source projects related to Snips https://snips.ai/.
Stars: ✭ 50 (+38.89%)
Mutual labels: kaldi
Kaldi Ioc++ Kaldi IO lib (static and dynamic).
Stars: ✭ 22 (-38.89%)
Mutual labels: kaldi
Vosk ServerWebSocket, gRPC and WebRTC speech recognition server based on Vosk and Kaldi libraries
Stars: ✭ 277 (+669.44%)
Mutual labels: kaldi
Montreal Forced AlignerCommand line utility for forced alignment using Kaldi
Stars: ✭ 490 (+1261.11%)
Mutual labels: kaldi
dropclass speakerDropClass and DropAdapt - repository for the paper accepted to Speaker Odyssey 2020
Stars: ✭ 20 (-44.44%)
Mutual labels: kaldi
Zamia SpeechOpen tools and data for cloudless automatic speech recognition
Stars: ✭ 374 (+938.89%)
Mutual labels: kaldi
kaldi-timit-sre-ivectorDevelop speaker recognition model based on i-vector using TIMIT database
Stars: ✭ 17 (-52.78%)
Mutual labels: kaldi
PykaldiA Python wrapper for Kaldi
Stars: ✭ 756 (+2000%)
Mutual labels: kaldi
Asr theory语音识别理论,论文和PPT
Stars: ✭ 344 (+855.56%)
Mutual labels: kaldi
Theano Kaldi RnnTHEANO-KALDI-RNNs is a project implementing various Recurrent Neural Networks (RNNs) for RNN-HMM speech recognition. The Theano Code is coupled with the Kaldi decoder.
Stars: ✭ 31 (-13.89%)
Mutual labels: kaldi
EspressoEspresso: A Fast End-to-End Neural Speech Recognition Toolkit
Stars: ✭ 808 (+2144.44%)
Mutual labels: kaldi
Awesome KaldiThis is a list of features, scripts, blogs and resources for better using Kaldi ( http://kaldi-asr.org/ )
Stars: ✭ 393 (+991.67%)
Mutual labels: kaldi
Speaker Verification task in Voxceleb1 dataset
This repository contains simple scripts for a training i-vector speaker recognition system on Voxceleb1[1] dataset using Kaldi. It was modified based on run.sh file on Kaldi/egs/sre10.
Requirement
How to use
- Move all files to {kaldi_root}/egs/sre10 folder
- Modify dataset directories and parameters in run.sh file to fit in your machine.
- Run run.sh file
Result
The 2048 component GMM-UBM and 600-dimensional i-vector extractor were trained using voxceleb1 training data for verification task. Training parameter is almost same compared to sre10 baseline on Kaldi egs.
GMM-2048 CDS eer : 15.39%
GMM-2048 LDA+CDS eer : 8.103%
GMM-2048 PLDA eer : 5.446%
Note
The Voxceleb1 dataset, a large-scale speaker identification dataset was published in 2017 with speaker embedding baseline[1] and reported i-vector shows 8.8% EER. The i-vector was extracted using 1024 component GMM-UBM, so the EER is fairly worse compared to the result above.
Reference
[1] A. Nagraniy, J. S. Chung, and A. Zisserman, “VoxCeleb: A large-scale speaker identification dataset,” in Interspeech, 2017, pp. 2616–2620.
Note that the project description data, including the texts, logos, images, and/or trademarks,
for each open source project belongs to its rightful owner.
If you wish to add or remove any projects, please contact us at
[email protected].