GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
Trelis
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
Trelis
Model Tree
Base
ARTPARK-IISc/whisper-large-v3-vaani-hindi
Fine-tuned
this model
Input Modalities
Audio
Output Modalities
Text
Supported Functionality
Dedicated Endpoints
Container
Available on Trelis Router (https://router.trelis.com/models):
POST https://router.trelis.com/api/v1/transcribe for programmatic access (requires an API key).Corpus WER (%, lower is better) under a script-safe indic-hindi normaliser (NFC + Indic
normalisation, keeps Devanagari matras/nuktas, strips punctuation; not the Whisper default, which strips
matras and inflates Devanagari WER). Compared against two leading commercial APIs: Sarvam (Saaras-v3)
and ElevenLabs Scribe-v2.
| Benchmark | whisper-hinglish-preview | Sarvam | Scribe-v2 | whisper-large-v3 | Vaani |
|---|---|---|---|---|---|
| CoSHE-500 (conversational CS) | 13.67 | 11.47 ᶜᵐ | 12.43 | 29.74 | 73.96 |
| cs-fleurs (read CS) | 10.19 | 16.47 ᶜᵐ | 7.57 | 33.92 | 34.12 |
| hiacc-adult (accented CS) | 12.73 | 14.44 ᶜᵐ | 16.98 | 28.53 | 60.09 |
| hiacc-child (accented CS) |
| Benchmark | whisper-hinglish-preview | Sarvam | Scribe-v2 | whisper-large-v3 | Vaani |
|---|---|---|---|---|---|
| Common Voice Hindi (cv-hi) | 12.86 | 12.40 | 13.44 | 30.82 | 14.48 |
| FLEURS-hi | 12.57 | 10.07 | 11.33 | 27.50 | 11.58 |
| Benchmark | whisper-hinglish-preview | Sarvam | Scribe-v2 | whisper-large-v3 | Vaani |
|---|---|---|---|---|---|
| FLEURS-en | 6.93 | 5.14 | 4.01 | 4.81 | 101.66 |
Bold = best on that row.
Like any Whisper model, specify the language when you transcribe.
python
from transformers import WhisperProcessor, WhisperForConditionalGenerationimport soundfile as sf, torchrepo = "Trelis/whisper-hinglish-preview"proc = WhisperProcessor.from_pretrained(repo)model = WhisperForConditionalGeneration.from_pretrained(repo, torch_dtype=torch.bfloat16).to("cuda").eval()audio, sr = sf.read("clip.wav") # 16 kHz monofeat = proc.feature_extractor(audio, sampling_rate=16000, return_tensors="pt").input_features.to("cuda", torch.bfloat16)# Hindi audio → force <|hi|> ; English audio → force <|en|>ids = proc.tokenizer.convert_tokens_to_idsprompt = [ids("<|startoftranscript|>"), ids("<|hi|>"), ids("<|transcribe|>"), ids("<|notimestamps|>")]out = model.generate(input_features=feat,decoder_input_ids=torch.tensor([prompt]).to("cuda"),max_new_tokens=440)print(proc.tokenizer.decode(out[0], skip_special_tokens=True))
Code-switched audio. The model uses a dedicated <|mixedcode|> marker/token for utterances that mix
Devanagari and Latin script. Insert it right after the language token, choosing the language token by the
dominant script of the utterance:
python
mc = proc.tokenizer("<|mixedcode|>", add_special_tokens=False).input_idsprompt = [ids("<|startoftranscript|>"), ids("<|hi|>"), *mc, ids("<|transcribe|>"), ids("<|notimestamps|>")]
<|hi|> / <|en|>) as shown above — standard Whisper usage — for the
reported quality.openai/whisper-large-v3.ARTPARK-IISc/whisper-large-v3-vaani-hindi,
a Vaani-fine-tuned Whisper-large-v3 from the Vaani project (ARTPARK @ IISc). We gratefully credit the
Whisper-Vaani model and the Vaani team.soketlabs/CoSHE-Eval (Soket Labs, CC-BY-NC-4.0).| 14.11 ᶜᵐ |
| 18.36 |
| 27.91 |
| 32.17 |