whisper onnxruntime all in one

main.py

"""
export model from https://github.com/microsoft/Olive/tree/main/examples/whisper
Audio should be 16khz mono
ffmpeg -i input.wav -ar 16000 -ac 1 -c:a pcm_s16le output.wav
"""
import numpy as np
import onnxruntime
from onnxruntime_extensions import get_library_path

audio_file = "single.wav" 
model = "whisper_cpu_int8_cpu-cpu_model.onnx"
with open(audio_file, "rb") as f:
    audio = np.asarray(list(f.read()), dtype=np.uint8)

inputs = {
    "audio_stream": np.array([audio]),
    "max_length": np.array([30], dtype=np.int32),
    "min_length": np.array([1], dtype=np.int32),
    "num_beams": np.array([5], dtype=np.int32),
    "num_return_sequences": np.array([1], dtype=np.int32),
    "length_penalty": np.array([1.0], dtype=np.float32),
    "repetition_penalty": np.array([1.0], dtype=np.float32),
}

options = onnxruntime.SessionOptions()
options.register_custom_ops_library(get_library_path())
session = onnxruntime.InferenceSession(model, options, providers=["CPUExecutionProvider"])
outputs = session.run(None, inputs)[0][0][0]
print(outputs)