Downloads · 30 days
9
15% of all-time downloads
lfm2-5-audio-jp-ss-jdsc-tts-v1
todo1111/lfm2-5-audio-jp-ss-jdsc-tts-v1
lfm2-5-audio-jp-ss-jdsc-tts-v1 is a text-to-speech model from todo1111. Use it when you need text read aloud. It is set up for liquid-audio. The card lists the license as other.
Fine-tuned LiquidAI/LFM2.5-Audio-1.5B-JP for single-speaker Japanese TTS on SS-JDSC.
All-time downloads
61
Public
Parameters
1.5B
3.6 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors3.6 GB · 100%
Parameter types
How the weights are stored.
BF161.5B · 100%
Model card
From the Hugging Face model README
lfm2-5-audio-jp-ss-jdsc-tts-v1
Fine-tuned LiquidAI/LFM2.5-Audio-1.5B-JP for single-speaker Japanese TTS on SS-JDSC.
Training
- Base model:
LiquidAI/LFM2.5-Audio-1.5B-JP - Run:
v1 - Prompt:
Perform TTS in japanese. - Data:
JDSC-Project/SS-JDSCtrain_13.0h single-speaker corpus - Context length: 384
- Training: 3000 steps, batch size 32 (~9.5 epochs)
- Checkpoint source: Modal Volume
ssjdsc-tts:/out/v1/final
Notes
This checkpoint is the TTS generator used to create synthetic SS-JDSC augmentation audio for ASR experiments.
The basic and hard variants in downstream experiments are produced by changing the input text generation, not by using separate TTS checkpoints.
Files
This repo contains the fine-tuned Liquid Audio checkpoint files copied from:
/vol/out/v1/final
- Task
- Text-to-Speech
- Library
- liquid-audio
- License
- other
- Created
- Jun 7, 2026
- Updated
- Jun 7, 2026