VoiceLab bundles: - lamejs MP3 encoder (LGPL-3.0) @breezystack/lamejs - mp4-muxer, webm-muxer (MIT) by Vanilagy - @elevenlabs/client (MIT), livekit-client (Apache-2.0), libsamplerate-js (BSD-2) - pdf.js (Apache-2.0) by Mozilla; fflate (MIT); Whisper speech-recognition models (MIT, downloaded at runtime) - kokoro-js (Apache-2.0), @huggingface/transformers (Apache-2.0), onnxruntime-web (MIT); Kokoro-82M model (Apache-2.0, downloaded at runtime) - Pocket TTS engine port (app/js/tts/pocket.js) follows the Apache-2.0 browser port by Kevin Knoedler (huggingface.co/spaces/KevinAHM/pocket-tts-web); Pocket TTS by Kyutai (code MIT, github.com/kyutai-labs/pocket-tts); model weights (CC-BY-4.0, kyutai/pocket-tts terms) are downloaded at runtime and not bundled - Chatterbox engine port (app/js/tts/chatterbox.js) follows the official ONNX inference scripts of Resemble AI Chatterbox (MIT, github.com/resemble-ai/chatterbox; ONNX exports ResembleAI/chatterbox-turbo-ONNX and onnx-community/chatterbox-multilingual-ONNX, MIT); model weights are downloaded at runtime - CosyVoice client (app/js/tts/cosyvoice.js) talks to the Apache-2.0 CosyVoice FastAPI server (github.com/FunAudioLLM/CosyVoice); no CosyVoice code or weights are bundled - OpenVoice engine (app/js/tts/openvoice.js) ports the tone color converter of MyShell OpenVoice V2 (MIT, github.com/myshell-ai/OpenVoice); ONNX exports TigreGotico/voiceclonnx-openvoice-v2 and Miyauel/openvoice-v2-onnx (MIT) are downloaded at runtime - Neural Server client (app/js/tts/neural.js) and tools/voicelab_neural_server.py call SWivid F5-TTS (code MIT, weights CC-BY-NC), Coqui XTTS v2 via idiap coqui-tts (code MPL-2.0, weights Coqui Public Model License), SparkAudio Spark-TTS (code Apache-2.0, weights CC-BY-NC-SA) and Fish Audio Fish Speech (Fish Audio Community License); no code or weights of these projects are bundled - GPT-SoVITS client (app/js/tts/gptsovits.js), the training-set slicer (app/js/docs/dataset.js, port of tools/slicer2.py) and tools/voicelab_gsv_bridge.py follow RVC-Boss GPT-SoVITS (MIT, github.com/RVC-Boss/GPT-SoVITS); no GPT-SoVITS code or weights are bundled VoiceLab is distributed under GPL-3.0-or-later.