first commit

.gitignore

@@ -0,0 +1,6 @@
+.env
+
+gen-lang-client-0065207637-eaf8e92995b6.json
+
+*.wav
+prompt.txt

gemini_call.py

@@ -0,0 +1,82 @@
+import google.generativeai as genai
+import os
+
+final_prompt = f'''Task:
+
+    Input Message:
+    Analyze the sentiment of each sentence in the provided conversation.
+    For each sentence, assign a sentiment score ranging from -10 (furious) to +10 (peaceful).
+    Generate a histogram of the conversation's sentences, where:
+    Negative numbers represent furious sentences.
+    Higher numbers represent more peaceful sentences.
+    At the end of the analysis, summarize the general sentiment of the conversation.
+    
+    Instructions for Sentiment Analysis:
+
+    Sentence-Level Sentiment Analysis:
+
+    For each sentence, determine its sentiment using the scale from -10 to +10.
+    -10: Extremely furious or hostile.
+    0: Neutral or balanced.
+    +10: Extremely peaceful or calm.
+    Example Sentences:
+
+    "I am so angry right now!" => Sentiment Score: -9 (furious)
+    "Everything is going wrong today." => Sentiment Score: -7 (frustrated)
+    "But, I guess there’s nothing I can do." => Sentiment Score: -3 (resigned but slightly peaceful)
+    "Maybe tomorrow will be better." => Sentiment Score: +3 (optimistic and calm)
+
+    Create a Histogram:
+
+    Based on the sentiment scores, plot a histogram where:
+    The x-axis represents the sentence number.
+    The y-axis represents the sentiment score.
+    The values should range from -10 (furious) to +10 (peaceful).
+    Example Histogram (hypothetical values):
+
+    Sentence 1: -9
+    Sentence 2: -7
+    Sentence 3: -3
+    Sentence 4: +3
+
+    Histogram:
+    | Sentence 1 | Sentence 2 | Sentence 3 | Sentence 4 |
+    |------------|------------|------------|------------|
+    |     -9     |     -7     |     -3     |     +3     |
+
+    Conclusion:
+
+    After analyzing all sentences, provide a conclusion about the overall sentiment of the conversation.
+    If the majority of sentences have a negative sentiment (below 0), the conversation is likely to be angry, frustrated, or tense.
+    If the majority of sentences have a positive sentiment (above 0), the conversation is peaceful or optimistic.
+    If the sentiments are mixed (both positive and negative), summarize the shift in mood throughout the conversation.
+
+    Example of Output:
+
+    Sentiment Scores:
+    "I am so angry right now!" => Sentiment Score: -9 (Furious)
+    "Everything is going wrong today." => Sentiment Score: -7 (Frustrated)
+    "But, I guess there’s nothing I can do." => Sentiment Score: -3 (Resigned)
+    "Maybe tomorrow will be better." => Sentiment Score: +3 (Optimistic)
+
+    Histogram:
+    Sentence 1: -9
+    Sentence 2: -7
+    Sentence 3: -3
+    Sentence 4: +3
+    (Plot: A simple histogram with y-axis ranging from -10 to +10, showing the corresponding sentiment values.)'''
+
+
+api_key = os.getenv("GOOGLE_API_KEY")
+genai.configure(api_key=api_key)
+model = genai.GenerativeModel("gemini-2.0-flash-exp" , system_instruction=final_prompt)
+
+def ask_gemini(prompt:str) ->str:
+    
+    
+    response = model.generate_content(prompt)
+    return response.text
+
+if __name__=="__main__":
+    response = model.generate_content("I hated every minute of my experience with you guys. You are straight garbage. But i love the way you look")
+    print(response.text)

sentiment_analysis.py

@@ -0,0 +1,17 @@
+from translation import get_transcription_from_sound
+from gemini_call import ask_gemini
+from dotenv import load_dotenv
+
+load_dotenv()
+
+def get_analysis(file_path)->str:
+    
+    transcript = get_transcription_from_sound(file_path)
+    
+    analysis = ask_gemini(transcript)
+    
+    return analysis
+
+if __name__ == "__main__":
+    file_path = "harvard.wav"
+    print(f'Analysis result {get_analysis(file_path)}')

streamlit_app.py

@@ -0,0 +1,70 @@
+from sentiment_analysis import get_analysis
+
+import streamlit as st
+import speech_recognition as sr
+import pyaudio
+import wave
+import time
+
+
+# Function to record the user's voice and save it as a .wav file
+def record_voice():
+    recognizer = sr.Recognizer()
+    mic = sr.Microphone()
+
+    # Set up the microphone and record the audio
+    with mic as source:
+        st.write("Listening...")
+        recognizer.adjust_for_ambient_noise(source)
+        audio = recognizer.listen(source)
+        st.write("Recording complete!")
+
+    # Save the audio to a .wav file
+    with open("recorded_audio.wav", "wb") as f:
+        with mic as source:
+            audio = recognizer.listen(source,timeout=5)
+            f.write(audio.get_wav_data())
+    
+    # Return the file name of the recorded audio
+    return "recorded_audio.wav", audio
+
+# Function to transcribe the audio file
+def transcribe_audio(audio):
+    recognizer = sr.Recognizer()
+
+    try:
+        # Recognize the speech using Google's speech recognition service
+        text = recognizer.recognize_google(audio)
+        st.write(f"Recognized text: {text}")
+        return text
+    except sr.UnknownValueError:
+        st.error("Sorry, I could not understand the audio.")
+    except sr.RequestError as e:
+        st.error(f"Error with the speech recognition service: {e}")
+    return None
+
+# Streamlit app setup
+st.title("Voice Chat App")
+
+st.sidebar.header("Controls")
+start_button = st.sidebar.button("Start Recording")
+
+if start_button:
+    st.write("Clicking this will start recording your voice...")
+    time.sleep(2)  # Pause for a moment before starting to record
+    audio_file, audio_data = record_voice()
+
+    # Saving and displaying the recorded audio
+    st.write(f"Audio saved as: {audio_file}")
+
+    # Transcribe the recorded audio
+    if audio_data:
+        transcription = transcribe_audio(audio_data)
+        if transcription:
+            st.text_area("Chat", value=transcription, height=200)
+        else:
+            st.write("Sorry, no transcription available.")
+    else:
+        st.write("No audio recorded.")
+
+

translation.py

@@ -0,0 +1,62 @@
+from google.cloud import speech
+import wave
+from pydub import AudioSegment
+
+def get_audio_properties(file_path):
+    """
+    Get sample rate and number of channels from the WAV file.
+    """
+    with wave.open(file_path, "rb") as wav_file:
+        sample_rate = wav_file.getframerate()
+        channels = wav_file.getnchannels()
+    return sample_rate, channels
+
+def convert_to_mono(input_path, output_path):
+    audio = AudioSegment.from_wav(input_path)
+    mono_audio = audio.set_channels(1)
+    mono_audio.export(output_path, format="wav")
+
+def transcribe_audio(file_path):
+    # Initialize the speech client
+    client = speech.SpeechClient()
+
+    # Get audio properties like sample rate and channels
+    sample_rate, channels = get_audio_properties(file_path)
+
+    # Open the audio file and read its content
+    with open(file_path, "rb") as audio_file:
+        audio_content = audio_file.read()
+
+    # Prepare the audio content for transcription
+    audio = speech.RecognitionAudio(content=audio_content)
+    config = speech.RecognitionConfig(
+        encoding=speech.RecognitionConfig.AudioEncoding.LINEAR16,
+        sample_rate_hertz=sample_rate,
+        language_code="en-US",
+        audio_channel_count=channels,
+        enable_separate_recognition_per_channel=(channels > 1)
+    )
+
+    # Call the Google Cloud Speech API for recognition
+    response = client.recognize(config=config, audio=audio)
+
+    # Concatenate the transcripts from all results into one string
+    concatenated_transcript = ". ".join([result.alternatives[0].transcript for result in response.results])
+
+    # Return or print the concatenated transcript
+    return concatenated_transcript
+
+def get_transcription_from_sound(file_path:str)->str:
+    output_path = "audio_mono.wav"
+    convert_to_mono(file_path,output_path)
+    
+    final_transcript = transcribe_audio(output_path)
+    
+    return final_transcript
+
+if __name__=="__main__":
+    file_path = "jackhammer.wav"
+    output_path = "audio_mono.wav"
+    convert_to_mono(file_path,output_path)
+    final_transcript = transcribe_audio(output_path)
+    print(final_transcript)