fix: retrieve mp3 bad request

.gitignore

@@ -160,3 +160,7 @@ cython_debug/
 #  and can be added to the global gitignore or merged into this file.  For a more nuclear
 #  option (not recommended) you can uncomment the following to ignore the entire idea folder.
 #.idea/
+
+
+*.mp3
+.DS_Store

README.md

@@ -1,6 +1,12 @@
 # Youtube-Whisper
 A simple Gradio app that transcribes YouTube videos by extracting audio and using OpenAI’s Whisper model for transcription. Paste a YouTube link and get the video’s audio transcribed into text.
 
+## Requirements
+
+- Conda installed (for managing environments)
+- Python 3.9 or above
+- **FFmpeg** installed (required for audio conversion)
+
 ## Installation
 
 ### Step 1: Clone the Repository
@@ -10,7 +16,31 @@ git clone https://github.com/danilotpnta/Youtube-Whisper.git
 cd Youtube-Whisper
 ```
 
-### Step 2: Create and Activate the Conda Environment
+### Step 2: Install FFmpeg
+
+You need FFmpeg for processing the audio. Install it based on your operating system:
+
+- **macOS**: Install FFmpeg via Homebrew:
+  ```bash
+  brew install ffmpeg
+  ```
+
+- **Ubuntu/Linux**: Install FFmpeg via apt:
+  ```bash
+  sudo apt update
+  sudo apt install ffmpeg
+  ```
+
+- **Windows**: 
+  - Download FFmpeg from the official website: [FFmpeg Download](https://ffmpeg.org/download.html).
+  - Extract the files and add the `bin` folder to your system’s PATH environment variable. For detailed instructions on adding FFmpeg to PATH, you can follow [this guide](https://www.geeksforgeeks.org/how-to-install-ffmpeg-on-windows/).
+
+Verify the installation by running:
+```bash
+ffmpeg -version
+```
+
+### Step 3: Create and Activate the Conda Environment
 
 To set up the environment using the provided `environment.yml` file:
 
@@ -24,7 +54,7 @@ Once the environment is created, activate it with:
 conda activate yt-whisper
 ```
 
-### Step 3: Run the App
+### Step 4: Run the App
 
 Once the environment is active, you can launch the Gradio app with:
 
@@ -36,8 +66,15 @@ This will start a local server for the app, and you can access it by visiting th
 
 ### Troubleshooting
 
-If you encounter any issues during installation, ensure that `pip` and `conda` are up to date:
+1. **FFmpeg Not Found**: 
+   If you see an error related to `ffmpeg not found`, ensure FFmpeg is installed and added to your system's PATH. You can also specify its location manually in the script by setting `ffmpeg_location`.
 
-```bash
-conda update conda
-pip install --upgrade pip
+2. **Pytube Errors**:
+   If you encounter issues with `pytube`, ensure you’re using the `yt-dlp` version and that your URL is correctly formatted.
+
+3. **Update Dependencies**:
+   Ensure that `pip` and `conda` are up to date:
+   ```bash
+   conda update conda
+   pip install --upgrade pip
+   ```

app.py

@@ -0,0 +1,51 @@
+import yt_dlp
+import whisper
+import gradio as gr
+import os
+
+# Function to download the audio from YouTube using yt-dlp
+def download_audio(url):
+    ydl_opts = {
+        'format': 'bestaudio/best',
+        'outtmpl': 'audio.%(ext)s',
+        'postprocessors': [{
+            'key': 'FFmpegExtractAudio',
+            'preferredcodec': 'mp3',
+            'preferredquality': '192',
+        }],
+    }
+
+    try:
+        with yt_dlp.YoutubeDL(ydl_opts) as ydl:
+            ydl.download([url])
+        audio_file = "audio.mp3"
+        return audio_file
+    except Exception as e:
+        return str(e)  # Return the error message for debugging
+
+# Function to transcribe the downloaded audio using Whisper
+def transcribe_audio(audio_path):
+    model = whisper.load_model("base")  # Use other models like "small", "medium", "large" if necessary
+    result = model.transcribe(audio_path)
+    return result['text']
+
+# Main function to integrate download and transcription
+def transcribe_youtube_video(youtube_url):
+    audio_path = download_audio(youtube_url)
+    if not os.path.exists(audio_path):  # Check if an error was returned
+        return f"Error: {audio_path}"  # Return the error message to the user
+    transcription = transcribe_audio(audio_path)
+    return transcription
+
+# Gradio interface setup using gradio.components
+interface = gr.Interface(
+    fn=transcribe_youtube_video,
+    inputs=gr.components.Textbox(label="YouTube URL"),
+    outputs=gr.components.Textbox(label="Transcription"),
+    title="YouTube Video Transcription",
+    description="Paste a YouTube video link to get the audio transcribed using Whisper."
+)
+
+# Launch the app
+if __name__ == "__main__":
+    interface.launch(share=True)  # Enables sharing with public link

environment.yml

@@ -5,8 +5,11 @@ channels:
 dependencies:
   - python=3.9
   - pip
+  - numpy<2  # Pinning NumPy to a version below 2.0 to avoid compatibility issues
   - pip:
-      - gradio==3.16.2
-      - pytube==12.1.0
+      - gradio==3.39.0  # Downgrade Gradio to work with Pydantic v1
+      - pytube==15.0.0
       - openai-whisper==20230314
-      - torch==2.0.1
+      - torch==2.0.1
+      - yt-dlp
+      - pydantic==1.10  # Use Pydantic v1 to avoid the incompatibility