modelapi/whisper-large-v3-fp16-ov-catalog

Whisper Large v3 automatic speech recognition model, a build of openai/whisper-large-v3 converted to OpenVINO™ IR with FP16 weights. It transcribes audio to text across many languages. Model weights are hosted in the source repository OpenVINO/whisper-large-v3-fp16-ov.

How to Use

1. Install required packages:

pip install optimum[openvino] "datasets<4" librosa soundfile --extra-index-url https://download.pytorch.org/whl/cpu

2. Run model inference:

from datasets import load_dataset
from transformers import AutoProcessor
from optimum.intel.openvino import OVModelForSpeechSeq2Seq

model_id = "OpenVINO/whisper-large-v3-fp16-ov"
processor = AutoProcessor.from_pretrained(model_id)
model = OVModelForSpeechSeq2Seq.from_pretrained(model_id)

dataset = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation", trust_remote_code=True)
sample = dataset[0]

input_features = processor(
    sample["audio"]["array"],
    sampling_rate=sample["audio"]["sampling_rate"],
    return_tensors="pt",
).input_features

outputs = model.generate(input_features)
print(processor.batch_decode(outputs)[0])

Legal information

The original model is distributed under the Apache-2.0 license. More details can be found in the original model card.