HyperAIHyperAI

Command Palette

Search for a command to run...

Speech Separation & Enhance

Isolating individual speech signals from mixed audio (separation) and improving speech quality by removing noise (enhancement).

Leaderboard

38 models total

Benchmarks

VoiceBank + DEMAND
CMGAN
Deep Noise Suppression (DNS) Challenge
Conv-TasNet-SNR
WHAMR!
Conv-TasNet DAE
EARS-WHAM
Schrödinger bridge (PESQ loss)
chime-3
Inter-Channel Conv-TasNet
VCTK
iEEG Clinical Dataset
IEEE
TIMIT
DNS challenge
lrs2
TDFNet-small
LRS3+DNS
LibriTTS Test-clean
LibriSpeech test-clean
LibriSpeech
DNS3
VoiceBank+DEMAND
3rd AVSEC dataset
spatialized WSJCAM0
TIMIT+MUSAN
TCD-TIMIT corpus (mixed-speech)
unseen speaker test set
Speech-Sound Event Separation
Speech Separation
WSJ0-SI 84+NOISEX-92
VBD
VibraVox (rigid in-ear microphone)
VibraVox (forehead accelerometer)
VibraVox (soft in-ear microphone)
VibraVox (temple vibration pickup)
VibraVox (throat microphone)
WHAM!
SepFormer
WHAMR_ext
Conv-TasNet DAE
WSJ0-2mix-extr
EARS-Reverb
SGMSE+
AudioMarkBench
AVDnR
AVED
AVSpeech + AudioSet
CHiME-4
Cocktail Party
DEMAND
Wave-U-Net
DnRv2
DnRv3
spatialized DNS challenge
DeFT-AN
EARS-WHAM!
easycom
MaxDI (Baseline)
GRID corpus (mixed-speech)
Audio-Visual concat-ref
LibriTTS Test-other
LJSpeech
MLS
MusicCaps
Sound Event Separation
SoundSpaces-Speech