GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
anak10thn
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
anak10thn
Model Tree
openai/whisper-large-v3-turbo
Input Modalities
Output Modalities
Supported Functionality
WER / CER (%), full test sets, lowercase + punctuation stripped:
| Test set | whisper-large-v3-turbo | whisper-large-v3-turbo-id |
|---|---|---|
| FLEURS id (684 utts) | 7.1 / 2.3 | 6.4 / 2.1 |
| Common Voice 17 id (3629 utts) | 10.1 / 3.5 | 7.7 / 2.5 |
Telephone / Opus / AAC channel + short context (500 CV clips, training-time eval at step 2000): 9.3 WER (vanilla turbo 10.8).
faster-whisper, int8_float16, Tesla T4, single stream, 4 s utterances: 298 (whisper-small-id: 106) ms per utterance.
Output is lowercase without punctuation.
model.safetensors (fp16) + tokenizer/config: transformersct2-int8/: CTranslate2 int8_float16 for faster-whisperpython
from huggingface_hub import snapshot_downloadfrom faster_whisper import WhisperModelpath = snapshot_download("anak10thn/whisper-large-v3-turbo-id", allow_patterns="ct2-int8/*")model = WhisperModel(f"{path}/ct2-int8", device="cuda", compute_type="int8_float16")segments, _ = model.transcribe("audio.wav", language="id", beam_size=1, vad_filter=True)print(" ".join(s.text for s in segments))python
from transformers import pipelineasr = pipeline("automatic-speech-recognition", model="anak10thn/whisper-large-v3-turbo-id", device=0)print(asr("audio.wav", generate_kwargs={"language": "indonesian", "task": "transcribe"})["text"])from transformers import pipelineasr = pipeline("automatic-speech-recognition", model="anak10thn/whisper-large-v3-turbo-id", device=0)print(asr("audio.wav", generate_kwargs={"language": "indonesian", "task": "transcribe"})["text"])