Explore @BUTSpeechFIT Open Source projects

BUT Speech@FIT (@BUTSpeechFIT)

BUTSpeechFIT

Stars
950
Global Org. Rank 15,881 (Top 6 %)
Registered about 5 years ago
Most used languages

Python
74.1 %

Jupyter Notebook
14.8 %

Shell
11.1 %
Location 🇨🇿 Czechia
Country Total Rank 22,674
Country Ranking

Jupyter Notebook
6

Shell
26

Python
27

VBx

Variational Bayes HMM over x-vectors diarization

x-vector-kaldi-tf

Tensorflow implementation of x-vector topology on top of Kaldi recipe

speakerbeam

Jupyter Notebook

EEND

DiariZen

A toolkit for speaker diarization.

Jupyter Notebook

AMI-diarization-setup

EEND_dataprep

DiaPer

MultiSV

MultiSV: scripts for data preparation

BrnoLM

A neural language modeling toolkit built on PyTorch

DVBx

Discriminative Training of VBx Diarization

ASR-hybrid-decoding

torch_msbg_mbstoi

Differentiable implementation of MSBG hearing loss model and MBSTOI intelligibility metric for Clarity Enhancement challenge.

safe_gpu

Avoids race condition when acquiring GPUs in exclusive mode

hystoc

Getting confidences from any end-to-end systems

cgmm_mvdr_online

Implementation of CGMM-MVDR beamforming used for Clarity challenge

Jupyter Notebook

vae_dolphin

diacorrect

Error correction back-end for speaker diarization

BaySMM

A Bayesian Multilingual Document Model

OOV-recovery-in-hybrid-ASR-system

beer

Jupyter Notebook

diarization_utils

TS-ASR-Whisper

vq-aud

CALLHOME_sublists

EENDEDA_VIB

wespeaker_ssl_public

Using Pre-trained SSL Transformer Models for Speaker Verification

tse_tutorial

speller

LAS-based E2E system training with speller

RAVDESS2Mix

wer_metric

Compute Word Error Rate