Releases · Sharrnah/whispering (original) (raw)

v1.3.18.10

v1.3.18.9

v1.3.18.6

v1.3.18.4

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 (5.0 GB):

Download Server:

• EU: https://eu2.contabostorage.com/bf1a89517e2643359087e5d8219c0c67:projects/whispering/whispering-tiger1.3.18.4_win.zip
• US: https://usc1.contabostorage.com/8fcf133c506f4e688c7ab9ad537b5c18:projects/whispering/whispering-tiger1.3.18.4_win.zip

Changelog (v1.3.18.4)

• [FEATURE] Add zonos hybrid model support
• [FEATURE] Add triton jit compiler support
• [FEATURE] Add microphone passthrough
• [TASK] Remove debug output in orpheus TTS
• [TASK] Add call to fix possible CUDNN_STATUS_SUBLIBRARY_VERSION_MISMATCH error
• [TASK] Send activity notification while speaking every 2 seconds
• [TASK] Print used cuDNN version on startup
• [TASK] Simplify realtime translation condition
• [TASK] Send timestamps setting to nemo_canary model
• [TASK] Optimize sentence splitting for kokoro tts
• [TASK] Update transformer library
• [TASK] Unify function arguments for kokoro
• [TASK] Add mamba_ssm dependency
• [TASK] Enhance Zonos TTS with queue, streaming improvements & text splitting
• [TASK] Update requirements to working versions
• [TASK] Phi4 code improvements
• [TASK] Additional cleanup of Phi-4 translation results
• [TASK] Check if cuda is available before loading triton backend for zonos tts
• [TASK] Add osc_force_activity_indication setting
• [TASK] Keep line-breaks when splitting osc messages due to length
• [TASK] Add returning of probabilities for language detections
• [TASK] Add language code converter
• [TASK] Save and load tts specific settings
• [BUGFIX] Add unidic dependency for kokoro
• [BUGFIX] Run kokoro tts on configured ai device
• [BUGFIX] Run orpheus tts on configured ai device
• [BUGFIX] Orpheus tts compatibility with newest transformers version
• [BUGFIX] Add missing libraries espeakng_loader and unidic_lite
• [BUGFIX] espeak usage with Zonos
• [BUGFIX] Use configured seed for streamed playback in zonos tts
• [BUGFIX] Implement Phi4 patch for transformers versions >= 4.50
• [BUGFIX] Correct torchaudio dependency for pytorch version
• [BUGFIX] Exit child processes on exit
• [BUGFIX] Fix text translation realtime sync
• [BUGFIX] Correctly save and load modified dictionaries in settings

Full Changelog: v1.3.17.3...v1.3.18.4

v1.3.17.3

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 (4.3 GB):

Download Server:

• EU: https://eu2.contabostorage.com/bf1a89517e2643359087e5d8219c0c67:projects/whispering/whispering-tiger1.3.17.3_win.zip
• US: https://usc1.contabostorage.com/8fcf133c506f4e688c7ab9ad537b5c18:projects/whispering/whispering-tiger1.3.17.3_win.zip

Changelog (v1.3.17.3)

• [FEATURE] Update pyTorch and flash-attn to NVIDIA 50x Blackwell supporting versions
• [FEATURE] Add support for flash and normal canary models
• [FEATURE] Add plugin settings reset function.
• [FEATURE] Add Orpheus TTS
• [FEATURE] Add audio streaming to Orpheus TTS without vllm
• [FEATURE] Add delayed start audio streaming
• [FEATURE] Implement Voxtral (Speech-to-Text + Text-Translation + LLM)
• [TASK] Remove debug output
• [TASK] cleanup transformer whisper code
• [TASK] Add flash attention 2 to transformer whisper
• [TASK] Update zonos library
• [TASK] call plugin_tts_after_audio event for streamed playback
• [TASK] Improve kokoro tts split_pattern
• [TASK] code cleanup
• [TASK] Add support for parakeet model
• [TASK] Implement BitsAndBytes for Transformer whisper again
• [TASK] Add large distilled v3.5 english model to faster whisper
• [TASK] Add setting to synchronize txt realtime with stt realtime active
• [TASK] Update transformers dependency
• [BUGFIX] internal event calling
• [BUGFIX] Fix canary code for new nemo library version
• [BUGFIX] Fix downloading MMS models
• [BUGFIX] Fix pyinstaller build spec

Full Changelog: v1.3.16.2...v1.3.17.3

v1.3.17.2-beta (CUDA 12.8)

v1.3.16.2

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 (4.3 GB):

Download Server:

• EU: https://eu2.contabostorage.com/bf1a89517e2643359087e5d8219c0c67:projects/whispering/whispering-tiger1.3.16.2_win.zip
• US: https://usc1.contabostorage.com/8fcf133c506f4e688c7ab9ad537b5c18:projects/whispering/whispering-tiger1.3.16.2_win.zip

Changelog (v1.3.16.2)

• [FEATURE] Add clear transcriptions function
• [FEATURE] select_textvalue and select_completion widgets
• [FEATURE] Add Zonos TTS
• [FEATURE] Add Phi-4 multimodal model (For STT, Translation, OCR)
• [FEATURE] Add function calling from Phi4 model
• [FEATURE] Add Kokoro TTS model
• [FEATURE] Add new OCR type (GOT OCR 2.0)
• [FEATURE] Add chat websocket message for multimodal LLM models like Phi-4mm
• [FEATURE] Add support for custom F5-TTS models
• [TASK] cache voice list for F5 TTS
• [TASK] Update desktop+ documentation
• [TASK] Allow streamed audio playback for TTS result
• [TASK] Display Torch version at startup
• [TASK] Update dependencies
• [TASK] Add download_model function to downloader
• [TASK] Add TTS stop call
• [TASK] Add streamed playback for F5-TTS, Kokoro TTS and Zonos TTS
• [TASK] uniffy OCR window capture over all OCR models
• [TASK] Encode image data after result sending
• [TASK] Add TTS audio normalization setting
• [BUGFIX] Streamed audio playback delaying last buffered audio frame
• [BUGFIX] Model loading for float16 only models in other precisions
• [BUGFIX] Make TTS model classes singletons
• [BUGFIX] Add additional punkt dependencies in build
• [BUGFIX] Fix osc message if LLM already supports transcription+translation
• [BUGFIX] Error if result object has no "text"
• [BUGFIX] Wav export with streamed playback enabled
• [BUGFIX] Make sure only one download per file is running
• [BUGFIX] Add missing dependency for Phi-4 model

Full Changelog: v1.3.15.4...v1.3.16.2

v1.3.15.4

v1.3.15.2

v1.3.15.1