SylReg-LM 7B
Model Details
Model Description
- Model type: Qwen2ForCausalLM
- Language(s) (NLP): English
- License: CC BY-NC-SA 4.0
- Finetuned from model: Qwen/Qwen2.5-7B
Model Sources
- Repository: Code
- Paper: arXiv:2607.04064
- Demo: Project page
How to Get Started with the Model
Use the code below to get started with the model.
git clone https://github.com/ryota-komatsu/speaker_disentangled_hubert.git
cd speaker_disentangled_hubert
sudo apt install git-lfs # for UTMOS
conda create -y -n py310 -c pytorch -c conda-forge python=3.10 pip=24.0 setuptools=81.0.0 faiss-gpu=1.13.2 uv sox
conda activate py310
export UV_PROJECT_ENVIRONMENT=$CONDA_PREFIX
uv pip install -r requirements/requirements.txt
sh scripts/setup.sh
import re
import torch
import torchaudio
from transformers import AutoModelForCausalLM, AutoTokenizer
from src.flow_matching import FlowMatchingWithBigVGan
from src.s5hubert import SylRegForSyllableDiscovery
wav_path = "/path/to/wav"
# download pretrained models from hugging face hub
encoder = SylRegForSyllableDiscovery.from_pretrained("ryota-komatsu/SylReg-Distill", device_map="cuda", dtype="auto")
decoder = FlowMatchingWithBigVGan.from_pretrained("ryota-komatsu/SylReg-Decoder", device_map="cuda", dtype="auto")
speechlm = AutoModelForCausalLM.from_pretrained("ryota-komatsu/SylReg-LM-7B", device_map="cuda", dtype="auto")
tokenizer = AutoTokenizer.from_pretrained("ryota-komatsu/SylReg-LM-7B")
# load a waveform
waveform, sr = torchaudio.load(wav_path)
waveform = torchaudio.functional.resample(waveform, sr, 16000)
# encode a waveform into syllabic units
outputs = encoder(waveform.to(encoder.device))
units = outputs[0]["units"] # [3950, 67, ..., 503]
# speech language modeling
text = "".join(f"<{unit}>" for unit in units)
input_ids = tokenizer(text, padding=True, return_tensors="pt").input_ids.to(speechlm.device)
generated_ids = speechlm.generate(input_ids=input_ids, do_sample=True, temperature=0.8)[0]
units = tokenizer.decode(generated_ids)
units = torch.tensor([int(unit) for unit in re.findall(r"<(\d+)>", units)], device=decoder.device)
# unit-to-speech synthesis
generated_speech = decoder(units.unsqueeze(0)).waveform.cpu()
Training Details
Training Data
| Hours | License | Provider | |
|---|---|---|---|
| LibriSpeech | 960 | CC BY 4.0 | V. Panayotov et al. |
| Libriheavy | 50,978 | public domain | W. Kang et al. |
| Emilia-Large | 4,447 | CC BY 4.0, CC BY-NC 4.0 | H. He et al. |
| People's Speech (clean, clean_sa) | 5,640 | CC-BY, CC-BY-SA | D. Galvez et al. |
| VoxPopuli | 543 | CC0-1.0 | C. Wang et al. |
| TinyStories | 27,810 | cdla-sharing-1.0 | R. Eldan et al. |
| Cosmopedia-v2 | 38,986 | odc-by | L. B. Allal et al. |
| Total | 129,364 |
Training Hyperparameters
- Training regime: bf16 mixed precision
- Training steps: 15k
- Batch size: 2,097,152 (=221) tokens
- Optimizer: AdamW(lr=0.0003, betas=(0.9, 0.95), weight_decay=0.01)
- Scheduler: warmup_stable_decay(num_warmup_steps=100, num_decay_steps=5000, min_lr_ratio=0.1)
Hardware
32 NVIDIA H100 GPUs
Citation
BibTeX:
@article{Komatsu_SylReg_2026,
author = {Komatsu, Ryota and Kawakita, Kota and Okamoto, Takuma and Shinozaki, Takahiro},
title = {Speaker-Disentangled Chunk-Wise Regression for Syllabic Tokenization},
year = {2026},
volume = {7},
journal = {IEEE Open Journal of Signal Processing},
pages = {800--808},
}