File size: 1,908 Bytes
ab58fb5 9d727e3 f3e24b8 b74cc3f ab58fb5 253b4e4 866ffb3 b74cc3f 3e7fc84 866ffb3 253b4e4 b03d0c5 4ebf6a9 b03d0c5 3277eaa 2040736 4ebf6a9 2040736 34b0634 2040736 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 |
import streamlit as st
import whisper
import os
from transformers import pipeline
from pydub import AudioSegment
def transcribe_audio(audiofile):
st.session_state['audio'] = audiofile
print(f"audio_file_session_state:{st.session_state['audio'] }")
#get size of audio file
audio_size = round(os.path.getsize(st.session_state['audio'])/(1024*1024),1)
print(f"audio file size:{audio_size}")
#determine audio duration
podcast = AudioSegment.from_mp3(st.session_state['audio'])
st.session_state['audio_segment'] = podcast
podcast_duration = podcast.duration_seconds
print(f"Audio Duration: {podcast_duration}")
st.info('Breaking podcast into 5 minute chunks.')
#break into 5 minute chunks
chunk_length_five_minutes = 5 * 60 * 1000
podcast_chunks = podcast[::chunk_length_five_minutes]
st.info('Transcribe')
#transcriptions = []
#for i, chunk in enumerate(podcast_chunks):
# chunk.export(f'output/chunk_{i}.mp4', format='mp4')
# following blogpost here: https://huggingface.co/blog/asr-chunking
transcribe_pipe = pipeline(model="facebook/wav2vec2-base-960h")
transcription = transcribe_pipe(audiofile, chunk_length_s=10, stride_length_s=(4, 2))
print(f"transcription: {transcription}")
return podcast_duration
st.markdown("# Podcast Q&A")
st.markdown(
"""
This helps understand information-dense podcast episodes by doing the following:
- Speech to Text transcription - using OpenSource Whisper Model
- Summarizes the episode
- Allows you to ask questions and returns direct quotes from the episode.
"""
)
if st.button("Process Audio File"):
transcribe_audio("marketplace-2023-06-14.mp3")
#audio_file = st.file_uploader("Upload audio copy of file", key="upload", type=['.mp3'])
# if audio_file:
# transcribe_audio(audio_file)
|