| import gradio as gr | |
| from transformers import pipeline | |
| pipe = pipeline(model="lfurman/whisper-tiny-en") | |
| def transcribe(audio): | |
| text = pipe(audio)["text"] | |
| return text | |
| with gr.Blocks() as demo: | |
| gr.Markdown("# Whisper Tiny FreeSound Audio Captioning") | |
| gr.Markdown("Upload an audio file for captioning using a fine-tuned Whisper tiny model.") | |
| with gr.Row(): | |
| audio_input = gr.Audio(sources="upload", type="filepath") | |
| text_output = gr.Textbox(label="Audio Caption") | |
| btn = gr.Button("Transcribe") | |
| btn.click(fn=transcribe, inputs=audio_input, outputs=text_output) | |
| if __name__ == "__main__": | |
| demo.launch() |