Credits
Clipabay stands on the work of many people who share their tools and sounds for free. Thank you.
Instrument sounds
- Salamander Grand Piano by Alexander Holm, licensed CC BY 3.0, via the Tone.js project.
- tonejs-instruments (guitars, bass, strings, winds, brass, harp, organ, harmonium, xylophone) collected and edited by Nicholas Brosowsky from public domain sources, licensed CC BY 3.0. Source.
AI models
- Vocal separation: Mel-Band RoFormer Duality (student model by musetric, from Aname's Mel-Band RoFormer Duality), Apache 2.0; Demucs (Meta), MIT, through demucs-web.
- Speech to text: Whisper (OpenAI), MIT. Translation: OPUS-MT (Helsinki-NLP), CC BY 4.0, and M2M100 (Meta), MIT.
- Voices: Kokoro (hexgrad), Apache 2.0; Piper voices (Rhasspy and contributors), each under its own license.
- Speaker detection: pyannote segmentation 3.0, MIT; WeSpeaker ResNet34 voice prints (trained on VoxCeleb), CC BY 4.0.
- Noise removal: DeepFilterNet3 (Hendrik Schröter), MIT or Apache 2.0, and RNNoise (Xiph.Org, Jean-Marc Valin), BSD.
- Slow motion: RIFE (hzwer), MIT. Upscaling: Real-ESRGAN (Xintao Wang), BSD 3-Clause. B-roll matching: CLIP (OpenAI), MIT. Voices in other languages turn text into speech sounds with espeak-ng through piper-phonemize, GPL 3.0.
- Background removal and face finding: MediaPipe (Google), Apache 2.0.
- These models run through Transformers.js (Hugging Face) and ONNX Runtime Web (Microsoft), both open source.
Stock media and sounds
Stock videos and photos come from Pixabay, and sound search uses Freesound. Each file keeps the license and credit shown next to it.
Open source code
MP4 and WebM files are written with mp4-muxer and webm-muxer by Vanilagy (MIT). MP3 files are made with lamejs (LGPL).