Switch transcription backend from Whisper to Mistral AI API

README.md

@@ -1,6 +1,6 @@
 # Audio Summary with Local LLM
 
-This tool is designed to provide a quick and concise summary of audio and video files. It supports summarizing content either from a local file or directly from YouTube. The tool uses Whisper for transcription and a local version of Llama3 (via Ollama) for generating summaries.
+This tool is designed to provide a quick and concise summary of audio and video files. It supports summarizing content either from a local file or directly from YouTube. The tool uses Mistral AI API for transcription and a local version of Llama3 (via Ollama) for generating summaries.
 
 > [!TIP]  
 > It is possible to change the model you wish to use.
@@ -10,7 +10,7 @@ This tool is designed to provide a quick and concise summary of audio and video
 
 - **YouTube Integration**: Download and summarize content directly from YouTube.
 - **Local File Support**: Summarize audio/video files available on your local disk.
-- **Transcription**: Converts audio content to text using Whisper.
+- **Transcription**: Converts audio content to text using Mistral AI API.
 - **Summarization**: Generates a concise summary using Llama3 (Ollama).
 - **Transcript Only Option**: Option to only transcribe the audio content without generating a summary.
 - **Device Optimization**: Automatically uses the best available hardware (MPS for Mac, CUDA for NVIDIA GPUs, or CPU).
@@ -23,6 +23,8 @@ Before you start using this tool, you need to install the following dependencies
 - [Ollama](https://ollama.com) for LLM model management
 - `ffmpeg` (required for audio processing)
 - [uv](https://docs.astral.sh/uv/getting-started/installation/) for package management
+- Mistral AI API key (free tier available)
+- Mistral AI API key (set as `MISTRAL_API_KEY` environment variable)
 
 ## Installation
 
@@ -39,6 +41,17 @@ uv sync
 source .venv/bin/activate  # On Windows: .venv\Scripts\activate
 ```
 
+### Mistral API Setup
+
+1. Sign up for a Mistral AI account and get your API key from the [Mistral AI platform](https://mistral.ai/)
+2. Set your API key as an environment variable:
+
+```bash
+export MISTRAL_API_KEY="your-api-key-here"
+```
+
+Or add it to your `.bashrc` or `.zshrc` file for persistent use.
+
 ### LLM Requirement
 
 [Download and install](https://ollama.com) Ollama to carry out LLM Management. More details about LLM models supported can be found on the Ollama [GitHub](https://github.com/ollama/ollama).
@@ -196,24 +209,21 @@ OLLAMA_MODEL = "llama3"
 WHISPER_MODEL = "openai/whisper-large-v2"
 ```
 
-#### Changing the Whisper Model
+#### Changing the Mistral Transcription Model
 
-To use a different Whisper model for transcription:
+To use a different Mistral model for transcription:
 
-1. Update the `WHISPER_MODEL` variable with one of these options:
-   - `"openai/whisper-tiny"` (fastest, least accurate)
-   - `"openai/whisper-base"` (faster, less accurate)
-   - `"openai/whisper-small"` (balanced)
-   - `"openai/whisper-medium"` (slower, more accurate)
-   - `"openai/whisper-large-v2"` (slowest, most accurate)
+1. Update the `MISTRAL_MODEL` variable with one of the available Mistral audio models:
+   - `"voxtral-mini-latest"` (default, good balance)
+   - `"voxtral-small-latest"` (smaller, faster)
 
 2. Example:
 
    ```python
-   WHISPER_MODEL = "openai/whisper-medium"  # A good balance between speed and accuracy
+   MISTRAL_MODEL = "voxtral-small-latest"  # Faster transcription
    ```
 
-For CPU-only systems, using a smaller model like `whisper-base` is recommended for better performance.
+The Mistral API handles all hardware acceleration automatically, so no device selection is needed.
 
 #### Changing the LLM Model