 Kaushik Ram Sadagopan
|
fb59ee0a49
Enable M4T vocoder inference on the GPU in fp16. (#151)
|
2 years ago |
 Yilin Yang
|
64c0e73ac0
Adding watermarked PretsselVocoderAgent (#149)
|
2 years ago |
 Anna Sun
|
8373db2ee5
Enable joint s2t + s2s output for demo (#146)
|
2 years ago |
|
Kaushik Ram Sadagopan
|
26bc428198
Clean up aligner assets, refactor aligner test to run on GPU with fp16 as well. (#148)
|
2 years ago |
|
Kaushik Ram Sadagopan
|
e568857c64
Fix all mypy issues in streaming, and some minor bugs in tree pipeline. (#147)
|
2 years ago |
 Abinesh Ramakrishnan
|
b5b98699c6
Tree Pipeline (#140)
|
2 years ago |
|
Yilin Yang
|
7d076869ef
update ProsodyUnitY checkpoint to fairseq2 format (#143)
|
2 years ago |
|
Yilin Yang
|
26a020227f
bump watermarked vocoder's max_seq_len to allow longer input (#145)
|
2 years ago |
 Ning
|
fad6308372
Fix medium model loading (#144)
|
2 years ago |
|
Abinesh Ramakrishnan
|
a1b5d918eb
Vocoder SimulEval Agent and getting online S2ST parity. (#135)
|
2 years ago |
|
Yilin Yang
|
8d49dd0450
small afterwards fix due to conda env break yesterday (#142)
|
2 years ago |
 Ilia Kulikov
|
78d6dac3a9
UnitY2 aligner for release (#112)
|
2 years ago |
 Can Balioglu
|
f9608d8cbd
Various nit fixes (#138)
|
2 years ago |
|
Anna Sun
|
3e673a7468
mixins come first (#137)
|
2 years ago |
|
Kaushik Ram Sadagopan
|
6fadf9e320
Introduce online unit decoder SimulEval agent. (#115)
|
2 years ago |
|
Anna Sun
|
d877073d7c
Bump simuleval version, override simuleval update_target to save memory (#133)
|
2 years ago |
 Tuan Tran
|
9f6ade6ee4
Make a public-facing watermarked vocoder (PretsselVocoder) (#97)
|
2 years ago |
|
Anna Sun
|
2ccf28ad24
[streaming] Port changes for streaming demo (#130)
|
2 years ago |
|
Yilin Yang
|
00118c21cc
Enabling 24khz vocoder for demo/OSS (#132)
|
2 years ago |
|
Kaushik Ram Sadagopan
|
7537081d50
Return durations from the variance adaptor. (#134)
|
2 years ago |
|
Abinesh Ramakrishnan
|
b1027e9858
Silero VAD Agent (#120)
|
2 years ago |
 Pierre Andrews
|
87e10d101b
mintox - Add option to consume pretranscribed text + log mintox for cloudwatch (#131)
|
2 years ago |
|
Kaushik Ram Sadagopan
|
5a2d61655f
Make unit_extractor configurable by dtype. (#128)
|
2 years ago |
|
Abinesh Ramakrishnan
|
bc88690d56
Ability to change tgt_lang dynamically during streaming inference. (#121)
|
2 years ago |
|
Kaushik Ram Sadagopan
|
5dd9722b8d
Introduce MMASpeechToTextDecoderAgent and related agents for online_text_decoder. (#113)
|
2 years ago |
 Ruslan Mavlyutov
|
c9f611a0b2
Fix inconsistency between model vocab info and associated tokenizers (inherit directly from the tokenizers) (#126)
|
2 years ago |
|
Yilin Yang
|
b9f101b2b7
Loose the PretsselModel test check by allowing one unit different b/t runs (#127)
|
2 years ago |
|
Kaushik Ram Sadagopan
|
239a9440a9
Offline w2v-bert encoder agent with parity. (#110)
|
2 years ago |
|
Kaushik Ram Sadagopan
|
521a374213
Online feature extractor SimulEval agent. (#107)
|
2 years ago |
|
Ruslan Mavlyutov
|
ca1ebf90ea
* Training recipees for M4T-nano/-micro\n* Adjustments to fairseq2/sc updates\n* Fixing MyPy warnings (#59)
|
2 years ago |
