Free Text-to-Speech Tool Adds Natural Voices Without Cloning
Now you can generate speech that sounds human without needing voice samples.
Deep Dive
Pocket TTS can now speak with predefined voices — no voice sample required. The kyutai/pocket-tts-without-voice-cloning weights ship predefined voices as the backbone KV state after the voice prompt: the converter turns each state back into prompt embeddings and stores them in the mmproj, the zeroed voice encoder is left out, and the "voice" field of /tts and --tts-voice select one of them. A speaker reference is rejected when the model has no voice encoder.
The shipped voice states aren't always the ones the
Key Points
- Pocket TTS now includes ready-to-use voices from Kyutai, so you can generate speech without recording a sample.
- The update adds checks to ensure voices sound correct and provides clear error messages.
- Developers can easily add multiple voices to their apps, potentially leading to more voice options in services you use.
Why It Matters
This makes it easier and cheaper to add natural-sounding voices to apps, potentially improving accessibility and entertainment.