| import gradio as gr, soundfile as sf, io | |
| from pydub import AudioSegment | |
| def audio_to_audiosegment(audio): | |
| sample_rate, audio_data = audio | |
| temp_wav = io.BytesIO() | |
| sf.write(temp_wav, audio_data, sample_rate, format='WAV') | |
| temp_wav.seek(0) | |
| [aud, vol, res] = [AudioSegment.from_file(temp_wav, format="wav"), [], ''] | |
| for i in range(int(len(aud) / 33)): | |
| frame = aud[i * 33 : (i + 1) * 33] | |
| vol.append(frame.dBFS) | |
| for v in vol: res += str([v < -50, -50 <= v and v < -30, -30 <= v and v < -20, -20 <= v and v < -10, -10 <= v].index(True) + 1) | |
| return res | |
| gr.Interface(fn=audio_to_audiosegment, inputs=gr.Audio(type="numpy"), outputs="text", title="口パク データ変換").launch() |