{webcamActive ? (
[truncated — 2641 more characters]
```
### vite.config.js
```javascript
import { defineConfig } from "vite";
import react from "@vitejs/plugin-react";
export default defineConfig({
plugins: [react()],
});
```
### index.html
```html
ConvoCoach
```
### test_speech_analysis.py
```python
import openai
import os
import re
import numpy as np
from dotenv import load_dotenv
import user_speech_to_text
# Load .env for API Key
load_dotenv()
api_key = os.getenv("OPENAI_API_KEY")
if not api_key:
raise ValueError("API key not found. Make sure '.env' exists and contains 'OPENAI_API_KEY'.")
# Initialize OpenAI client
client = openai.OpenAI(api_key=api_key)
#TODO: Add text speech feedback (num words in response/num_words)
def generate_avatar_response(transcribed_text):
"""
Uses OpenAI API to analyze speech and provide conversational feedback while keeping the topic in mind.
"""
prompt = f"""
Imagine you are a **vocal coach assistant** designed to help a wide variety of users improve their speech clarity while keeping the conversation flowing.
Your task is to:
1️⃣ **Analyze the user's speech transcript** and look for:
- **Unfinished thoughts** (sentences that end abruptly).
- **Repeated words** (signs of stuttering or hesitation).
- **Filler words** (uh, um, like, you know, actually, basically, so).
- **Long pauses** (indicated by "...").
2️⃣ **Continue the conversation naturally** by responding in a way that keeps the user's topic in mind.
3️⃣ **Provide subtle, encouraging feedback** in a conversational manner—avoid sounding too robotic or overly critical.
**Rules for Your Response:**
- Be **supportive and friendly**.
- Your response should be **2-3 short sentences**.
- First, **engage with what the user was trying to say**.
- Then, **gently provide feedback** to help them improve.
- Most importantly, only give feedback **when necessary**. If speech was good and coherent, compliment/commend them.
**User's Speech Transcript:**
"{transcribed_text}"
Based on this transcript, respond as if you were talking to them in real-time. Keep the conversation flowing while subtly helping them improve.
"""
response = client.chat.completions.create(
model="gpt-4-turbo",
messages=[{"role": "user", "content": prompt}],
temperature=0.6, # Balances structure and variation
max_tokens=200 # Ensures concise responses
)
return response.choices[0].message.content.strip()
# Transcript (Test Case)
def get_response():
audio_file = user_speech_to_text.record_audio()
transcript = user_speech_to_text.transcribe_audio(audio_file)
transcript_text = transcript.get("text", "").strip()
if not transcript_text:
print("⚠️ No valid transcription received. Exiting...")
exit()
# Get AI avatar response
avatar_response = generate_avatar_response(transcript_text)
print(avatar_response)
return avatar_response
```
### user_speech_to_text.py
```python
import os
import time
import wave
import assemblyai as aai
from dotenv import load_dotenv
# Load API Key
load_dotenv(".env")
aai.settings.api_key = os.getenv("ASSEMBLYAI_API_KEY")
if not aai.settings.api_key:
raise ValueError("AssemblyAI API key is missing. Make sure it's set in .env1!")
# Audio Configuration
RATE = 16000 # Sample rate
CHANNELS = 1 # Mono
RECORD_SECONDS = 15 # Duration to record
def record_audio():
"""Records audio from the microphone and saves it as a WAV file."""
import pyaudio
filename = f"user_audios/recorded_audio_{int(time.time())}.wav"
p = pyaudio.PyAudio()
stream = p.open(format=pyaudio.paInt16,
channels=CHANNELS,
rate=RATE,
input=True,
frames_per_buffer=1024)
print(f"Recording for {RECORD_SECONDS} seconds...")
frames = []
for _ in range(0, int(RATE / 1024 * RECORD_SECONDS)):
data = stream.read(1024)
frames.append(data)
print("Recording complete.")
stream.stop_stream()
stream.close()
p.terminate()
os.makedirs("user_audios", exist_ok=True)
with wave.open(filename, 'wb') as wf:
wf.setnchannels(CHANNELS)
wf.setsampwidth(p.get_sample_size(pyaudio.paInt16))
wf.setframerate(RATE)
wf.writeframes(b''.join(frames))
return filename # Return the filename for processing
def transcribe_audio(audio_file):
"""
Transcribes the given audio file and returns the transcript + sentiment analysis.
:param audio_file: Path to the audio file to transcribe.
:return: A dictionary containing transcription text and sentiment analysis results.
"""
transcriber = aai.Transcriber()
config = aai.TranscriptionConfig(sentiment_analysis=True)
transcript = transcriber.transcribe(audio_file, config)
try:
os.remove(audio_file)
except Exception as e:
print(f"Error deleting file: {e}")
return {
"text": transcript.text,
"sentiment_analysis": [
{
"text": result.text,
"sentiment": result.sentiment,
"confidence": result.confidence,
"start": result.start,
"end": result.end
}
for result in transcript.sentiment_analysis
]
}
# Allow this script to be run standalone for testing
if __name__ == "__main__":
audio_file = record_audio()
print("Transcribing and analyzing sentiment...")
result = transcribe_audio(audio_file)
print("\nTranscription:")
print(result["text"])
print("\nSentiment Analysis Results:")
for entry in result["sentiment_analysis"]:
print(f"Text: {entry['text']}")
print(f"Sentiment: {entry['sentiment']}")
print(f"Confidence: {entry['confidence']}")
print(f"Timestamp: {entry['start']} - {entry['end']}\n")
```
### avatar.py
```python
import requests
import time
from dotenv import load_dotenv
import os
import test_speech_analysis as chat_response
load_dotenv() # Load variables from .env file
api_key = os.getenv('AVATAR_API_KEY')
if api_key is None:
raise ValueError("AVATAR_API_KEY environment variable is not set")
# URL of the avatar image
# source_url = 'https://media-hosting.imagekit.io//4078ec62a6824c90/screenshot_1739688772692.png?Expires=1834296773&Key-Pair-Id=K2ZIVPTIP2VGHC&Signature=LwWvBbL2shd0AqMt~XyVwUwxey2AGKgPPrMEAql3lULNTvG8mF~Urcxhpy7KvW2Cy2LL1VWU69UBsOXc5SWU9FgsvvVz03ggSbtVtApbitZ2eh7uKwiv0rzsvhTFCwmJBxPpiU9D-CdY~ynDhlHodH~xE8wpd2WYAzdChBhLW2QfeEPDs3zP12LXeZFnO0-05Xco0LTnwQOUK12ZDUsIZLtkWIln6Ooqkb9R-BjBDKT9zGnw5IY5D2JHWrTiJOLX3H0W0Ti65BcXYrbYiILndV2xVM-aIfVPoCtz2bYrSPswkG5yPYDErl79jApaQCvIc3mjZBDvCpEjC4u2U-7xGw__'
# Henok
source_url = 'https://media-hosting.imagekit.io//b9a94f2580c647be/screenshot_1739696990595.png?Expires=1834304991&Key-Pair-Id=K2ZIVPTIP2VGHC&Signature=QoHsh~Mtks6TznmRGXzCjBoPPts-fRoXzjB~3arufkhEKO5eRo2GTIV8teM8PxseAVlYJJcV3am5~397h0K0-CPSzmD6UnkDcvJfpCPoYPzovU3hhWqXbGKD4Q~8YY5GcLwCuYbxiGAThLVqUgonT8tY2EuuE07YQTyM9XUHqJ5OmLauk9ikO~k6bkk1H75DecK0rMMI0CO3oHLs72fvE4LLNzvgIXiTmyzuDC0JxmvdZGsA6oZUghuLxtCGZr8DcvKNGlM~rlCVeGVTHWgl6RluUxrfdxfKXg9jLj7x0GlPia3M~VSkS9MW8C8jnCZJzNXN4msVV5iJh~bT5Tyfdg__'
# Text you want the avatar to speak
text_input = chat_response.get_response()
print("received response from text_speech_analysis")
# Endpoint to create a talk
create_talk_url = 'https://api.d-id.com/talks'
# Headers for the request
headers = {
'Authorization': f'Basic {api_key}',
'Content-Type': 'application/json'
}
# Data payload for the request
data = {
'source_url': source_url,
'script': {
# "provider": {
# "type": "amazon",
# "voice_id": "Joey"
# },
'type': 'text',
'input': text_input
}
}
response = requests.post(create_talk_url, headers=headers, json=data)
if response.status_code == 201:
talk_id = response.json().get('id')
print(f'Talk created successfully with ID: {talk_id}')
else:
print(f'Failed to create talk: {response.status_code} - {response.text}')
exit()
# Endpoint to retrieve the generated video
get_talk_url = f'https://api.d-id.com/talks/{talk_id}'
# Polling to check the status of the video generation
while True:
response = requests.get(get_talk_url, headers=headers)
if response.status_code == 200:
talk_data = response.json()
if talk_data.get('status') == 'done':
result_url = talk_data.get('result_url')
print(f'Video generated successfully: {result_url}')
break
elif talk_data.get('status') == 'error':
print('Error in video generation.')
break
else:
print('Video is being processed...')
else:
print(f'Failed to retrieve talk: {response.status_code} - {response.text}')
break
time.sleep(5) # Wait for 5 seconds before polling again
```
[8 more indexed source files omitted to keep this export small. The full file list is in the Codebase structure section above.]