TaSTT.git/TaSTT-Whisper, branch master Free self-hosted STT for VRChat. https://git.yummers.dev/TaSTT.git/atom?h=master 2023-06-27T00:21:59+00:00 Scrub out old C++-based Whisper code 2023-06-27T00:21:59+00:00 yum yum.food.vr@gmail.com 2023-06-27T00:21:59+00:00 urn:sha1:694756a96a6109cd79a77221dd4e40638ff55b82 No longer used. Begin integrating faster-whisper 2023-04-24T03:52:36+00:00 yum yum.food.vr@gmail.com 2023-04-24T03:52:36+00:00 urn:sha1:b4bb6524652e0f76834ca26a4afa232855ca1348 This is a much faster, lower-VRAM reimplementation of Whisper in Python. Early testing is extremely promising: fast transcription speed, extremely low resource usage (CPU/RAM/VRAM), high accuracy. Fix _socket module not found issue 2023-03-21T22:02:29+00:00 yum yum.food.vr@gmail.com 2023-03-21T21:28:46+00:00 urn:sha1:656d7c2092545b18d981acfac000c73fb2128e4a Need python310._pth, specifically 'import site' line, for embedded python + pip to get along. Expose more C++ whisper parameters in GUI 2023-03-08T22:59:57+00:00 yum yum.food.vr@gmail.com 2023-03-08T22:59:57+00:00 urn:sha1:b6dc76afb4f76a8b0eaa8f821c557cd83f69daf4 Expose decode method, beam search parameters, and voice activity detection parameters in GUI. * Remove WhisperCPP::Init(), do it on launch instead. * Add float support to ConfigMarshal Animator generation and dumping mics no longer hang GUI 2023-03-06T00:26:18+00:00 yum yum.food.vr@gmail.com 2023-03-06T00:26:18+00:00 urn:sha1:058cd57ed2d18d6e14408881c3229fab7edbadb1 Do these in a std::future. * SetAffinityMask() now returns a value on all control paths Bugfix: fix use-after-free in GetMicsImpl 2023-02-28T08:34:10+00:00 yum yum.food.vr@gmail.com 2023-02-28T08:34:10+00:00 urn:sha1:f65b93aa6f0a5b7ddd069cd1f50f142029e4a5e5 * Plumb beam search params into whisper cpp implementation (currently broken) Improve behavior around VAD segmentation events 2023-02-27T04:12:36+00:00 yum yum.food.vr@gmail.com 2023-02-27T04:12:36+00:00 urn:sha1:d96851f716b9e18d827ba7795a343dbf3cf529c4 Use forked Whisper implementation which has tweaks to reduce dropped words around the beginning VAD segments. * Retain audio after VAD segmentation events