Spaces:
Sleeping
Sleeping
| import json | |
| import requests | |
| from datetime import datetime | |
| import time | |
| import traceback | |
| API_URL = "https://api-inference.huggingface.co/models/" | |
| def date_now(): | |
| return datetime.now().strftime("%Y-%m-%d %H:%M:%S") | |
| def record_opt(msg): | |
| return f"{date_now()} {msg}\n" | |
| def speech_recognize(audio, model_name, hf_token, opt): | |
| opt += record_opt("Transcription starts ...") | |
| yield "Transcribing, please wait..", opt | |
| start = time.monotonic() | |
| with open(audio, "rb") as f: | |
| data = f.read() | |
| try: | |
| url = API_URL + model_name | |
| print(f">>> url is {url}") | |
| headers = {"Authorization": f"Bearer {hf_token}"} | |
| response = requests.request("POST", url, headers=headers, data=data) | |
| text = json.loads(response.content.decode("utf-8")) | |
| print(f">>> text is {text}") | |
| text = text['text'] | |
| except: | |
| text = f"Transcription failed：\n{traceback.format_exc()}" | |
| cost = time.monotonic() - start | |
| opt += record_opt(f"Transcription ends, time consuming{cost:.3f}s") | |
| yield text, opt | |
| import gradio as gr | |
| with gr.Blocks() as demo: | |
| gr.HTML("""<h2 align="center">Automatic Speech Recognition (OpenAI Whisper with Inference API)</h2>""") | |
| with gr.Row(): | |
| gr.Markdown( | |
| """🤗 Call the huggingface API and use the OpenAI Whisper model for speech recognition, which can also be called speech to text（Speech to Text, STT） | |
| 👉 The purpose is to practice using the Gradio Audio component and explore using the Huggingface Inference API | |
| > 💡Tip: You need to fill in the Huggingface token to call the Huggingface Inference API | |
| """ | |
| ) | |
| with gr.Row(): | |
| with gr.Column(): | |
| audio = gr.Audio(source="microphone", type="filepath") | |
| model_name = gr.Dropdown( | |
| label="Select model", | |
| choices=[ | |
| "openai/whisper-large-v3", | |
| "openai/whisper-large-v2", | |
| "openai/whisper-large", | |
| "openai/whisper-medium", | |
| "openai/whisper-small", | |
| "openai/whisper-base", | |
| "openai/whisper-tiny", | |
| ], | |
| value="openai/whisper-large-v3", | |
| ) | |
| hf_token = gr.Textbox(label="Huggingface token") | |
| with gr.Column(): | |
| output = gr.Textbox(label="Transcription results") | |
| operation = gr.Textbox(label="Component operation history") | |
| audio.start_recording( | |
| lambda x: x + record_opt("Start recording ..."), | |
| inputs=operation, outputs=operation | |
| ) | |
| audio.play( | |
| lambda x: x + record_opt("Play recording"), | |
| inputs=operation, outputs=operation | |
| ) | |
| audio.pause( | |
| lambda x: x + record_opt("Pause playback"), | |
| inputs=operation, outputs=operation | |
| ) | |
| audio.stop( | |
| lambda x: x + record_opt("Stop play"), | |
| inputs=operation, outputs=operation | |
| ) | |
| audio.end( | |
| lambda x: x + record_opt("Finished playing"), | |
| inputs=operation, outputs=operation | |
| ) | |
| audio.stop_recording(speech_recognize, inputs=[audio, model_name, hf_token, operation], outputs=[output, operation]) | |
| demo.queue(max_size=4, concurrency_count=4) | |
| demo.launch() | |