Liquid AILFM2.5
LFM2.5-Audio-1.5B-JP
This 1.5B speech model supports Japanese speech-to-speech conversation.
What the release provides
It combines an LFM2.5 backbone, a FastConformer input encoder and an audio-token generator. Interleaved generation serves conversation; sequential generation supports transcription and speech synthesis.LFM2.5-Audio-1.5B-JP official model card
How to use it well
Practical guidance. Check natural turn endings, personal names and mixed Japanese-English speech. Evaluate conversational delay separately from transcript accuracy.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | LiquidAI/LFM2.5-Audio-1.5B-JP |