import streamlit as st import moviepy.editor as mp import speech_recognition as sr from pydub import AudioSegment import tempfile import os import io from transformers import pipeline import numpy as np import wave import matplotlib.pyplot as plt # Function to convert video to audio def video_to_audio(video_file): # Load the video using moviepy video = mp.VideoFileClip(video_file) # Extract audio audio = video.audio temp_audio_path = tempfile.mktemp(suffix=".mp3") # Write the audio to a file audio.write_audiofile(temp_audio_path) return temp_audio_path # Function to convert MP3 audio to WAV def convert_mp3_to_wav(mp3_file): # Load the MP3 file using pydub audio = AudioSegment.from_mp3(mp3_file) # Create a temporary WAV file temp_wav_path = tempfile.mktemp(suffix=".wav") # Export the audio to the temporary WAV file audio.export(temp_wav_path, format="wav") return temp_wav_path # Function to transcribe audio to text def transcribe_audio(audio_file): # Initialize recognizer recognizer = sr.Recognizer() # Load the audio file using speech_recognition audio = sr.AudioFile(audio_file) with audio as source: audio_data = recognizer.record(source) try: # Transcribe the audio data to text using Google Web Speech API text = recognizer.recognize_google(audio_data) return text except sr.UnknownValueError: return "Audio could not be understood." except sr.RequestError: return "Could not request results from Google Speech Recognition service." # Function to perform emotion detection using Hugging Face transformers def detect_emotion(text): # Load emotion detection pipeline emotion_pipeline = pipeline("text-classification", model="j-hartmann/emotion-english-distilroberta-base", return_all_scores=True) # Get the emotion predictions result = emotion_pipeline(text) # Extract the emotion with the highest score emotions = {emotion['label']: emotion['score'] for emotion in result[0]} return emotions # Function to visualize audio waveform def plot_waveform(audio_file): with wave.open(audio_file, 'r') as w: signal = np.frombuffer(w.readframes(w.getnframes()), dtype=np.int16) plt.figure(figsize=(10, 4)) plt.plot(signal) plt.title("Audio Waveform") plt.xlabel("Sample") plt.ylabel("Amplitude") st.pyplot(plt) # Streamlit app layout st.title("Video and Audio to Text Transcription with Emotion Detection and Visualization") st.write("Upload a video or audio file to convert it to transcription, detect emotions, and visualize the audio waveform.") # Create tabs to separate video and audio uploads tab = st.selectbox("Select the type of file to upload", ["Video", "Audio"]) if tab == "Video": # File uploader for video uploaded_video = st.file_uploader("Upload Video", type=["mp4", "mov", "avi"]) if uploaded_video is not None: # Save the uploaded video file temporarily with tempfile.NamedTemporaryFile(delete=False) as tmp_video: tmp_video.write(uploaded_video.read()) tmp_video_path = tmp_video.name # Add an "Analyze Video" button if st.button("Analyze Video"): with st.spinner("Processing video... Please wait."): # Convert video to audio audio_file = video_to_audio(tmp_video_path) # Convert the extracted MP3 audio to WAV wav_audio_file = convert_mp3_to_wav(audio_file) # Transcribe audio to text transcription = transcribe_audio(wav_audio_file) # Show the transcription st.text_area("Transcription", transcription, height=300) # Emotion detection emotions = detect_emotion(transcription) st.write(f"Detected Emotions: {emotions}") # Plot the audio waveform st.subheader("Audio Waveform Visualization") plot_waveform(wav_audio_file) # Store transcription and audio file in session state st.session_state.transcription = transcription # Store the audio file as a BytesIO object in memory with open(wav_audio_file, "rb") as f: audio_data = f.read() st.session_state.wav_audio_file = io.BytesIO(audio_data) # Cleanup temporary files os.remove(tmp_video_path) os.remove(audio_file) # Check if transcription and audio file are stored in session state if 'transcription' in st.session_state and 'wav_audio_file' in st.session_state: # Provide the audio file to the user for download st.audio(st.session_state.wav_audio_file, format='audio/wav') # Add download buttons for the transcription and audio # Downloadable transcription file st.download_button( label="Download Transcription", data=st.session_state.transcription, file_name="transcription.txt", mime="text/plain" ) # Downloadable audio file st.download_button( label="Download Audio", data=st.session_state.wav_audio_file, file_name="converted_audio.wav", mime="audio/wav" ) elif tab == "Audio": # File uploader for audio uploaded_audio = st.file_uploader("Upload Audio", type=["wav", "mp3"]) if uploaded_audio is not None: # Save the uploaded audio file temporarily with tempfile.NamedTemporaryFile(delete=False) as tmp_audio: tmp_audio.write(uploaded_audio.read()) tmp_audio_path = tmp_audio.name # Add an "Analyze Audio" button if st.button("Analyze Audio"): with st.spinner("Processing audio... Please wait."): # Convert audio to WAV if it's in MP3 format if uploaded_audio.type == "audio/mpeg": wav_audio_file = convert_mp3_to_wav(tmp_audio_path) else: wav_audio_file = tmp_audio_path # Transcribe audio to text transcription = transcribe_audio(wav_audio_file) # Show the transcription st.text_area("Transcription", transcription, height=300) # Emotion detection emotions = detect_emotion(transcription) st.write(f"Detected Emotions: {emotions}") # Plot the audio waveform st.subheader("Audio Waveform Visualization") plot_waveform(wav_audio_file) # Store transcription in session state st.session_state.transcription_audio = transcription # Store the audio file as a BytesIO object in memory with open(wav_audio_file, "rb") as f: audio_data = f.read() st.session_state.wav_audio_file_audio = io.BytesIO(audio_data) # Cleanup temporary audio file os.remove(tmp_audio_path) # Check if transcription and audio file are stored in session state if 'transcription_audio' in st.session_state and 'wav_audio_file_audio' in st.session_state: # Provide the audio file to the user for download st.audio(st.session_state.wav_audio_file_audio, format='audio/wav') # Add download buttons for the transcription and audio # Downloadable transcription file st.download_button( label="Download Transcription", data=st.session_state.transcription_audio, file_name="transcription_audio.txt", mime="text/plain" ) # Downloadable audio file st.download_button( label="Download Audio", data=st.session_state.wav_audio_file_audio, file_name="converted_audio_audio.wav", mime="audio/wav" )