v1.3.15.4

Important:

This requires a lot of configuration if run directly. Recommended way is to use UI Application: https://github.com/Sharrnah/whispering-ui which downloads this automatically.

---

Standalone Release File (3.1 GB):

Download Server:

• EU: https://eu2.contabostorage.com/bf1a89517e2643359087e5d8219c0c67:projects/whispering/whispering-tiger1.3.15.4_win.zip

• US: https://usc1.contabostorage.com/8fcf133c506f4e688c7ab9ad537b5c18:projects/whispering/whispering-tiger1.3.15.4_win.zip

Changelog (v1.3.15.3)

• [FEATURE] Add get_last_generation methods for TTS
• [TASK] Add Greek F5 TTS Model
• [TASK] Add silence after segments of F5-TTS generation
• [TASK] F5 processing estimate for multi-segments
• [TASK] Update libraries + fix for nltk
• [TASK] Add pyctcdecode library
• [TASK] Add normalization to F5 TTS
• [TASK] unified tts event call
• [BUGFIX] Channel error on MME Audio API with Silero
• [BUGFIX] websocket disconnect on receiving generated TTS raw audio
• [BUGFIX] TTS Model download not starting on model change

Full Changelog: v1.3.15.2...v1.3.15.4

v1.3.15.2

Important:

This requires a lot of configuration if run directly. Recommended way is to use UI Application: https://github.com/Sharrnah/whispering-ui which downloads this automatically.

---

Standalone Release File (3.1 GB):

Download Server:

• EU: https://eu2.contabostorage.com/bf1a89517e2643359087e5d8219c0c67:projects/whispering/whispering-tiger1.3.15.2_win.zip

• US: https://usc1.contabostorage.com/8fcf133c506f4e688c7ab9ad537b5c18:projects/whispering/whispering-tiger1.3.15.2_win.zip

Changelog (v1.3.15.2)

• [TASK] Add setting only_no_speech_threshold_for_segments
• [TASK] Select random download url on fallback code
• [TASK] Add multilingual detection per segment for faster_whisper
• [BUGFIX] typo in setting transcription_auto_save_continuous_text
• [BUGFIX] Loading models with bfloat precisions
• [BUGFIX] Interleave support for more than 2 channels
• [BUGFIX] Custom model loading (faster whisper)
• [BUGFIX] fixed distilled-large-v3 model for faster-whisper

Full Changelog: v1.3.15.1...v1.3.15.2

v1.3.15.1

Important:

This requires a lot of configuration if run directly. Recommended way is to use UI Application: https://github.com/Sharrnah/whispering-ui which downloads this automatically.

---

Standalone Release File (3.1 GB):

Download Server:

• EU: https://eu2.contabostorage.com/bf1a89517e2643359087e5d8219c0c67:projects/whispering/whispering-tiger1.3.15.1_win.zip

• US: https://usc1.contabostorage.com/8fcf133c506f4e688c7ab9ad537b5c18:projects/whispering/whispering-tiger1.3.15.1_win.zip

Changelog (v1.3.15.1)

• [FEATURE] Update F5-TTS + add new F5-TTS language models (English, Chinese, French, German, Italian, Japanese, Spanish, Russian, Vietnamese, Malaysian)
• [FEATURE] Add option to change TTS volume
• [BUGFIX] Fix order of model loading
• [BUGFIX] Use configured F5-TTS compute device
• [BUGFIX] BigVGAN loading for F5-TTS
• [BUGFIX] module import order

Full Changelog: v1.3.14.8...v1.3.15.1

v1.3.14.8

Important:

This requires a lot of configuration if run directly. Recommended way is to use UI Application: https://github.com/Sharrnah/whispering-ui which downloads this automatically.

---

Standalone Release File (3.1 GB):

Download Server:

• EU: https://eu2.contabostorage.com/bf1a89517e2643359087e5d8219c0c67:projects/whispering/whispering-tiger1.3.14.8_win.zip

• US: https://usc1.contabostorage.com/8fcf133c506f4e688c7ab9ad537b5c18:projects/whispering/whispering-tiger1.3.14.8_win.zip

Changelog (v1.3.14.8)

• [FEATURE] Add F5 TTS
• [FEATURE] Add option to translate to more than one target language
• [FEATURE] Add OSC Server to synchronize with VRChat Mute state
• [FEATURE] Add support to load a user custom model
• [TASK] Add reload voices event
• [TASK] Update dependencies
• [TASK] Initialize TTS after UI connected
• [TASK] Only send source + translation if both actually exist
• [TASK] remove direct-ml for linux
• [TASK] Add large-v3-turbo model for faster-whisper
• [TASK] Open playback audio device directly with detected informations instead of trying multiple options
• [TASK] return audio segments in faster whisper
• [TASK] additional translation improvements
• [TASK] Upadate ctranslate library
• [BUGFIX] use defined exclude_client for BroadcastMessage
• [BUGFIX] Add possible stream playback fix
• [BUGFIX] Add linux build portaudio dependency
• [BUGFIX] Return correct download status on fallback download
• [BUGFIX] Error if invalid F5/E5 model is requested

Full Changelog: v1.3.14.6...v1.3.14.8

v1.3.14.6

Important:

This requires a lot of configuration if run directly. Recommended way is to use UI Application: https://github.com/Sharrnah/whispering-ui which downloads this automatically.

---

Standalone Release File (3.2 GB):

Download Server:

• EU: https://eu2.contabostorage.com/bf1a89517e2643359087e5d8219c0c67:projects/whispering/whispering-tiger1.3.14.6_win.zip

• US: https://usc1.contabostorage.com/8fcf133c506f4e688c7ab9ad537b5c18:projects/whispering/whispering-tiger1.3.14.6_win.zip

Changelog (v1.3.14.6)

• [FEATURE] Add stt_processing plugin function
• [TASK] Return detected language from transformer whisper
• [TASK] Update transformers library
• [BUGFIX] NLLB200 transformers based implementation
• [BUGFIX] Calculate correct chunk size for VAD v3 model (Fixes #26)
• [BUGFIX] vad_frames_per_buffer validity check
• [BUGFIX] API of M4T and NLLB-200 models with newer transformers library version
• [BUGFIX] disabled VAD error "KeyError: 'plugins'"

Full Changelog: v1.3.14.5...v1.3.14.6

v1.3.14.5

Important:

This requires a lot of configuration if run directly. Recommended way is to use UI Application: https://github.com/Sharrnah/whispering-ui which downloads this automatically.

---

Standalone Release File (3.2 GB):

Download Server:

• EU: https://eu2.contabostorage.com/bf1a89517e2643359087e5d8219c0c67:projects/whispering/whispering-tiger1.3.14.5_win.zip

• US: https://usc1.contabostorage.com/8fcf133c506f4e688c7ab9ad537b5c18:projects/whispering/whispering-tiger1.3.14.5_win.zip

Changelog (v1.3.14.5)

• [FEATURE] Direct-ML support. (Should allow to run many AI models on any DirectX 12 comatible GPU, including Intel and AMD)
• [TASK] Update dependencies
• [BUGFIX] Return original text in case TranslateLanguage thinks a text-translator is active but fails.
• [BUGFIX] voice markers reusing the initial audio data every time.

Full Changelog: v1.3.14.4...v1.3.14.5

v1.3.14.4

Important:

This requires a lot of configuration if run directly. Recommended way is to use UI Application: https://github.com/Sharrnah/whispering-ui which downloads this automatically.

---

Standalone Release File (3.1 GB):

Download Server:

• EU: https://eu2.contabostorage.com/bf1a89517e2643359087e5d8219c0c67:projects/whispering/whispering-tiger1.3.14.4_win.zip

• US: https://usc1.contabostorage.com/8fcf133c506f4e688c7ab9ad537b5c18:projects/whispering/whispering-tiger1.3.14.4_win.zip

Changelog (v1.3.14.4)

• [FEATURE] Write TTS result to file directly if path is provided.
• [TASK] Enable 'thread_per_transcription' by default again.
• [TASK] Show traceback on plugin error.
• [TASK] play_audio supporting bytes, torch.Tensor or numpy array
• [TASK] Add frozendict library (used by ChatTTS plugin)

Full Changelog: v1.3.14.2...v1.3.14.4

v1.3.14.2

Important:

This requires a lot of configuration if run directly. Recommended way is to use UI Application: https://github.com/Sharrnah/whispering-ui which downloads this automatically.

---

Standalone Release File (3.1 GB):

Download Server:

• EU: https://eu2.contabostorage.com/bf1a89517e2643359087e5d8219c0c67:projects/whispering/whispering-tiger1.3.14.2_win.zip

• US: https://usc1.contabostorage.com/8fcf133c506f4e688c7ab9ad537b5c18:projects/whispering/whispering-tiger1.3.14.2_win.zip

Changelog (v1.3.14.2)

• [FEATURE] Support for audio with more than 2 channels.
• [FEATURE] Add MMS STT model
• [FEATURE] clipboard image OCR support
• [FEATURE] Add select_audio widget for Plugins
• [FEATURE] Add textfield widget type
• [FEATURE] Add Speaker diarization class (experimental)
• [FEATURE] Add noisereduce algorythm
• [TASK] Improve streamed audio playback
• [TASK] use romaji setting for translation requests
• [TASK] Update ignorelist
• [TASK] add playback hook, simplify buffer size setting
• [TASK] Separation of audio processing for recording
• [TASK] Add get languages plugin method
• [TASK] Update dependencies
• [TASK] remove downloaded zip renaming
• [TASK] Add multiple file hash check utility function
• [TASK] Send loading message over stdout instead of websocket
• [TASK] Add plugin name to plugin errors
• [TASK] Upgrade dependencies + VAD model to v5
• [BUGFIX] catch plugin exceptions to not break whole application
• [BUGFIX] Fix possible process management error if process could not be run
• [BUGFIX] error on modified value in websocket message
• [BUGFIX] streamed playback of dynamic chunk size
• [BUGFIX] tagged streamed playback
• [BUGFIX] buffer element size calculation.
• [BUGFIX] Wait for resampling until full chunk is ready for streamed playback
• [BUGFIX] resample_audio function on gpu tensors, reshaping audio data
• [BUGFIX] Faster whisper handling of non avialable precision model files
• [BUGFIX] plugin on_*_call calls not returning anything.

Full Changelog: v1.3.13.1...v1.3.14.2

v1.3.13.1

Important:

This requires a lot of configuration if run directly. Recommended way is to use UI Application: https://github.com/Sharrnah/whispering-ui which downloads this automatically.

---

Standalone Release File (3.1 GB):

Download Server:

• EU: https://eu2.contabostorage.com/bf1a89517e2643359087e5d8219c0c67:projects/whispering/whispering-tiger1.3.13.1_win.zip

• US: https://usc1.contabostorage.com/8fcf133c506f4e688c7ab9ad537b5c18:projects/whispering/whispering-tiger1.3.13.1_win.zip

Changelog (v1.3.13.1)

• [FEATURE] Add seamlessM4T v2 model
• [FEATURE] Add Wav2Vec Bert2.0 STT Models
• [FEATURE] Add TextCorrection model (for Wav2Vec Bert2.0 models)
• [FEATURE] Add Whisper using Transformer library
• [FEATURE] Add NVIDIA NeMo Canary STT model
• [FEATURE] Add streaming overlay03 with 2 columns
• [FEATURE] plugin event methods
• [TASK] Load settings from Profile folder
• [TASK] Support more datatypes in audio processing methods
• [TASK] use pyaudio pool for audio streamer playback
• [TASK] Update libraries
• [TASK] Add annotated-types lib to build
• [TASK] Switch to CUDA 12.1
• [TASK] split pytorch requirements
• [TASK] Add triton for windows again
• [TASK] Change default vad_frames_per_buffer value
• [TASK] Add TTS playback over html to streaming overlay03
• [TASK] Make Whisper Voice Marker class a singleton
• [TASK] Process transcription in single thread by default
• [BUGFIX] use filename from provided url instead of last redirect
• [BUGFIX] Audio processing without VAD
• [BUGFIX] Multiprocess tasks running main python code
• [BUGFIX] loading whisper large-v3 in different precisions
• [BUGFIX] download whisper large-v3 when set to float32
• [BUGFIX] allow downloads without checksum

Full Changelog: v1.3.12.2...v1.3.13.1

v1.3.12.2

Standalone Release File (3.1 GB):

Download Server:

• EU: https://eu2.contabostorage.com/bf1a89517e2643359087e5d8219c0c67:projects/whispering/whispering-tiger1.3.12.2_win.zip

• US: https://usc1.contabostorage.com/8fcf133c506f4e688c7ab9ad537b5c18:projects/whispering/whispering-tiger1.3.12.2_win.zip

Changelog (v1.3.12.2)

• [FEATURE] Add Whisper V3 Support
• [FEATURE] Add Whisper Distilled Support
• [FEATURE] Add Option to write continuesly transcriptions to file (transcription_auto_save_continous_text setting)
• [FEATURE] Add Option to write the audio file of each final transcription (transcription_save_audio_dir setting)
• [FEATURE] Add buffered streamed audio playback
• [TASK] Replaced Icon
• [TASK] Replaced download library
• [TASK] Add grpcio library
• [TASK] Update omegaconf
• [TASK] Made DeepFilterNet model class a singleton

Full Changelog: v1.3.12.1...v1.3.12.2