GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
atx-labs
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Run this model inference with full control and performance in your environment.
Model Details
Model Provider
atx-labs
Model Tree
unsloth/orpheus-3b-0.1-ft
Input Modalities
Output Modalities
Supported Functionality
unsloth/orpheus-3b-0.1-ft)r=64, lora_alpha=128, rsLoRA)hi)The model was evaluated on a benchmark dataset of 210 Hindi evaluation prompts categorized into three core domains:
All synthesized 24 kHz Hindi audio samples were transcribed using AI4Bharat IndicConformer Multilingual ASR (ai4bharat/indic-conformer-600m-multilingual) to measure Word Error Rate (WER) and Character Error Rate (CER).
| Category | Prompt Count | Word Error Rate (WER) | Character Error Rate (CER) |
|---|---|---|---|
| Hindi Complex | 60 | 0.1657 | 0.0401 |
| Hindi Normal | 100 | 0.2913 | 0.0579 |
| Hindi Numbers | 50 | 0.4589 | 0.2885 |
| Overall Average |
"एक सौ पचास") rather than raw numeric digits (e.g., "१५०").Install the required dependencies:
bash
pip install torch transformers snac soundfile
generate_speech)python
import os, torch, soundfile as sffrom transformers import AutoModelForCausalLM, AutoTokenizerrepo_id = "atx-labs/voice-hindi-female"hf_token = os.environ.get("HF_TOKEN")tokenizer = AutoTokenizer.from_pretrained(repo_id, token=hf_token)model = AutoModelForCausalLM.from_pretrained(repo_id, trust_remote_code=True, torch_dtype=torch.bfloat16, token=hf_token).to("cuda").eval()# Synthesize speech directlywaveform = model.generate_speech("नमस्ते, आपका स्वागत है।", tokenizer)# Save 24 kHz WAV audiosf.write("output_hi.wav", waveform, 24000)print("✅ Audio generated and saved to output_hi.wav!")
transformers Pipelinepython
import os, torch, soundfile as sffrom transformers import pipelinerepo_id = "atx-labs/voice-hindi-female"hf_token = os.environ.get("HF_TOKEN")# Initialize TTS pipelinetts = pipeline("text-to-speech", model=repo_id, trust_remote_code=True, device=0, token=hf_token)# Synthesize speechoutput = tts("नमस्ते, आपका स्वागत है।")# Save 24 kHz WAV audiosf.write("output_hi.wav", output["audio"], output["sampling_rate"])print("✅ Audio generated and saved to output_hi.wav!")
Vyom TTS is released as part of OpenXFI, an open-source Indic AI initiative by ATX Labs. OpenXFI builds and openly releases foundational models, benchmarks, and datasets for Indian languages — with the goal of making high-quality Indic AI accessible to researchers, developers, and institutions across the country.
Explore the wider release, including our Indic evaluation benchmark and open datasets, at openxfi.com.
If you use this model in your research or applications, please cite both the model and the OpenXFI initiative:
bibtex
@misc{vyomttsxfi2026,title = {Vyom TTS: Open Indic Text-to-Speech Models},author = {Joshi, Mrugaja and Pathak, Guruprasad},year = {2026},publisher = {Hugging Face},howpublished = {\url{https://huggingface.co/atx-labs}},note = {Released as part of OpenXFI by ATX Labs}}
bibtex
@misc{openxfi2026,title = {OpenXFI: Shaping India's AI, Together},author = {{ATX Labs}},year = {2026},url = {https://openxfi.com}}
We also encourage citing the upstream work this model builds on — the Orpheus base model and the SNAC neural audio codec.
| 0.2953 |
| 0.1077 |