GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
LyngualLabs
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
LyngualLabs
Model Tree
openai/whisper-small
Input Modalities
Output Modalities
Supported Functionality
Scored against the plain-fine-tuned control with identical normalization. Language tags are stripped from both reference and hypothesis before computing WER/CER, so the transcription comparison is apples-to-apples; the tags are scored separately as LID.
| Metric | Tag-injection (this model) | Plain baseline |
|---|---|---|
| WER — tone-aware | 16.81% | 16.73% |
| WER — tone-insensitive | 14.04% | 13.86% |
| CER — tone-aware | 6.25% | 6.38% |
| CER — tone-insensitive | 5.02% | 5.10% |
| Per-word LID — accuracy | 99.54% | — |
| Per-word LID — macro-F1 | 99.54% | — |
Adding inline tags matches the plain model's WER (within run-to-run noise; CER is
actually a touch better) while delivering 99.54% per-word language identification for
free. For reference, the reported whisper-small-yoruba baseline is 20.76% WER; this
project's plain 5-epoch fine-tune reaches 16.73%.
The tag-vs-plain A/B was run identically on three model families:
| Model | WER tone-aware (plain → tag) | Per-word LID (tag) |
|---|---|---|
| Omnilingual CTC 300M | 33.57 → 32.45 | 98.04% |
| Omnilingual LLM 1.63B | 16.15 → 16.29 | 99.55% |
| Whisper-small (this) | 16.73 → 16.81 | 99.54% |
Inline language-tag injection costs ~nothing on WER (−1.1 to +0.1 absolute) and gives 98–99.5% free per-word LID on CTC, LLM, and Whisper alike.
language_tags are grouped
into contiguous same-language spans and wrapped: <yo> … </yo> <en> … </en>.add_tokens(..., special_tokens=False)
so they are ordinary vocabulary items that survive decode(skip_special_tokens=True)
(unlike control tokens, which would be dropped). Embeddings are resized to match.Seq2SeqTrainer, language conditioning yoruba,
task transcribe.NFC → lowercase → strip Unicode punctuation/symbol categories (P/S). Two variants:
The gap between the two measures how well the model handles Yoruba tone marking.
python
from transformers import WhisperForConditionalGeneration, WhisperProcessorimport torch, librosarepo = "LyngualLabs/yecs-asr-whisper-lid"proc = WhisperProcessor.from_pretrained(repo)model = WhisperForConditionalGeneration.from_pretrained(repo, torch_dtype=torch.bfloat16).to("cuda").eval()audio, _ = librosa.load("utterance.wav", sr=16000) # 16 kHz monofeats = proc.feature_extractor(audio, sampling_rate=16000, return_tensors="pt").input_featuresids = model.generate(feats.to("cuda", torch.bfloat16),language="yoruba", task="transcribe", max_new_tokens=225)print(proc.tokenizer.batch_decode(ids, skip_special_tokens=True)[0])# -> '<en> ... </en> <yo> ... </yo>'
Strip the tags for a plain transcript, or parse the spans to recover per-word language.
openai/whisper-small · 5 epochs · lr 1e-5 · bf16 · effective batch 32 · 1×H200.yoruba, task transcribe; forced_decoder_ids=None,
suppress_tokens=[]; best checkpoint by dev WER.afroscale_ai_cmu_africa/yecs-lid, run whisper-tag).Research on Yoruba–English code-switching ASR and per-word spoken language identification.
yo, en); other languages are out of scope.tag-injection/)Apache-2.0 (inherited from openai/whisper-small). YECS corpus terms apply to the
training data.