Liquid AI open-sources LiquidCodec, an on-device audio codec faster than same-size rivals
JosephJacks_ · x · 2026-11-13
- Liquid AI released LiquidCodec, its in-house streaming audio codec, now available on Hugging Face with a blog detailing architecture, training recipe, and results.
- It powers the recently launched LFM3-Audio-2B and LFM3-Omni-2B on-device multimodal models, converting speech to tokens and back.
- The 74M-parameter codec matches similar-sized models on reconstruction quality and compression while beating them on speech generation and token-to-waveform decoding speed, built for low-precision on-device inference.
- Evolution: earlier models used the open-source Mimi codec (1.1 kbps, 96M params); LFM2.5-Audio swapped only the detokenizer for 8x faster decoding, and LiquidCodec is now co-designed with the models end to end.
Related event: Liquid AI Open-Sources 74M-Parameter Audio Codec LiquidCodec(4 posts)→
More from Infra
- Perplexity's Mac app shows a local models option, hinting at offline capability — dnlkwk · 2026-11-13
- 2-bit DeepSeek V4.1 Flash runs a coding-agent seat on two 2016 V100s: a 12-hour real-work trial — a4mula · 2026-11-13
- IBM QDC 2026 Unveils Nighthawk r2 Quantum Chip and Fault-Tolerant Dev Tooling — MJBiercuk · 2026-11-13
- Why Jensen Huang Splits From US Frontier Labs on Open Source and Export Controls — morganlai · 2026-11-13
- SGLang-Neuron v0.1.0 open-sourced: SGLang serving stack now runs on AWS Trainium — ying11231 · 2026-11-13
- Used GPU prices on eBay ranked by $/GB: P100 at $10.9/GB, V100 servers nearly doubled in weeks — MzCWzL · 2026-11-13