AI

Kitten-TTS 2: New Open-source SOTA 1.58 bit TTS (CPU-only with voice cloning, multilingual, emotion control)

Read full article at source ↗

Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 22/10/2026, 06:47:25 UTC.

Briefings

Kitten-TTS 2 is a new open-source TTS model released by KittenML, featuring the world's first 1.58-bit ternary weights. It runs in real-time on CPU with zero-shot voice cloning and multilingual support, offering a lightweight alternative to ElevenLabs for GPU-poor environments. The model uses 1.7B parameters and supports 20+ languages with expressive emphasis control.

Key points

• The model uses ternary weights where every linear weight is one of three values (-1, 0, +1) with a group size of 128 for scales.
• It achieves zero-shot voice cloning from less than 5 seconds of reference audio while maintaining expressivity and accent information without fine-tuning.

Context

Released by KittenML, Kitten-TTS 2 represents a significant advancement in open-source TTS technology, offering a super-efficient frontier model that runs locally on consumer hardware.

Limitations

Evidence is limited to the article's summary of features; specific performance benchmarks, training data sizes, or deployment metrics are not provided.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 22/10/2026, 06:48:38 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeigkz66dv75c7env3ejwzq4pq6k6zzia6nwsdreeu4kgokoo5utfcu