Pocket TTS: Clone Any Voice in 5s on CPU
gvij · reddit · 2026-07-06
Kyutai's Pocket TTS is a streaming language model with roughly 100 million parameters. It uses the Mimi neural codec to generate audio tokens, enabling zero-shot voice cloning from just 5 seconds of reference audio without needing a GPU. It is MIT-licensed and easily installed via pip. The author compared it against three CPU TTS models: Kokoro, Supertonic, and Inflect-Nano. Across 180 timed tests, its latency remained nearly constant regardless of text length (RTF 0.69-0.76) and supported streaming output. Although it was the slowest of the four, it offered the most unique features.
Related event: Kyutai Open-Sources Pocket-TTS for CPU-Based Voice Cloning(4 posts)→
More from Multimodal
- ComfyUI Krea2 Prompt Styler Node Revived in a Fork With a Massive Style Library — Winter_unmuted · 2026-10-27
- Single HTML file generates retro poster designs in-browser with zero LLM calls — Afinetheorem · 2026-10-27
- A Redditor Used Copilot to Imagine and Archive Their Own '80s/'90s Winter Streetwear Concepts — Valuable-Sky2134 · 2026-10-27
- Transkun open-source piano transcription model nails MIDI, sustain events included — alexanderchen · 2026-10-27
- fal adds FLUX 3 Image Klein LoRA training: 10 reference images, 3 LoRAs per request — OdinLovis · 2026-10-27
- Guy turns his girlfriend's selfie into a playable AI game ad about popping pimples — CurieuxExplorer · 2026-10-27