Open-source credits
ReadAloud builds on the work of open-source projects. Thank you to their authors. Licences are the ones the projects publish; follow the links for the full terms.
Models and software
| Project | Used for | Licence |
|---|---|---|
| Kokoro-82M | Text-to-speech model | Apache License 2.0 |
| misaki | Text-to-phoneme front end for speech synthesis | Apache License 2.0 |
| Piper | Text-to-speech engine | GNU GPL v3.0 |
| eSpeak NG | Pronunciation (phonemisation) for speech synthesis | GNU GPL v3.0 |
| Chatterbox | Voice cloning from a reference recording | MIT |
| Whisper large-v3-turbo | Speech to text and the transcription step of dubbing | MIT |
Datasets
CC BY 4.0 requires attribution: CML-TTS, LibriTTS-R and Multilingual LibriSpeech are credited to their authors at the links below.
| Dataset | Used for | Licence |
|---|---|---|
| CML-TTS | Training data for Spanish dubbing voices | CC BY 4.0 |
| LibriTTS-R | Base data for Spanish dubbing voices | CC BY 4.0 |
| Multilingual LibriSpeech (MLS) | Training data for French dubbing voices | CC BY 4.0 |