Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/hay/wikiaudify
Generate audio summaries of Wikipedia articles using OpenAI and ElevenLabs
https://github.com/hay/wikiaudify
Last synced: about 2 months ago
JSON representation
Generate audio summaries of Wikipedia articles using OpenAI and ElevenLabs
- Host: GitHub
- URL: https://github.com/hay/wikiaudify
- Owner: hay
- License: mit
- Created: 2024-11-03T21:23:05.000Z (2 months ago)
- Default Branch: main
- Last Pushed: 2024-11-03T21:35:18.000Z (2 months ago)
- Last Synced: 2024-11-03T22:23:26.232Z (2 months ago)
- Language: Python
- Size: 9.82 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# wikiaudify
> Generate audio summaries of Wikipedia articles using OpenAI and ElevenLabs
By [Hay Kranen](http://www.haykranen.nl)
## Introduction
This was a hackathon project made during the [Wikimedia NL Mini Hackathon 2024](https://nl.wikimedia.org/wiki/Mini_Hackathon_November_2024) to generate audio summaries like the ones from [NotebookLM](https://blog.google/technology/ai/notebooklm-audio-overviews/). Obviously it's not as good as that one, but it makes quite enjoyable fun short audio conversations.
Here's an example of a generated audio summary about the article on [Grilled cheese](https://en.wikipedia.org/wiki/Grilled_cheese).
https://github.com/user-attachments/assets/59149473-146d-4c11-8b6d-c97e1c0eb01d
And you can find a transcription here.
## Install
What you'll need:
* An [OpenAI API key](https://openai.com/api/)
* An [ElevenLabs API key](https://elevenlabs.io/api)
* Python 3.13+ (it probably works with older versions too, but no guarantees)
There is an option to use a local LLM ([like Ollama](https://ollama.com/)) but i didn't get very good results, but you could try to make it work!
To use this script:
1. Clone this repo
```bash
git clone https://github.com/hay/wikiaudify.git
```
2. Make a virtual environment and install the `requirements.txt`
```bash
python -m venv env
source env/bin/activate
pip install -r requirements.txt
```
3. Copy the `example-config.toml` to a new file (e.g. `test.toml`) and fill in your API keys and other details
4. Try running it from the command line like this:
```bash
python generate.py -a "Grilled_Cheese" -q "At what temperature will my cheese melt?" -c test.toml
```
Note that the Wikipedia article you give with the `-a` option should have underscores, e.g. the path in the URL of the article.
By default this will generate two files in the root of this project: a `summary.mp3` containing the summary and a `summary.txt` with a transcription.
## Troubleshooting
If you add the `-v` (verbose) flag `audio2text` will give much more debug information.
## All options
You'll get this when doing `python generate.py -h`
```bash
usage: generate.py [-h] [-a ARTICLE] [-c CONFIG] [-na] [-nt] [-o OUT]
[-ot OUT_TRANSCRIPT] [-q QUESTION] [-v]
Generate an audio summary of a Wikipedia article
options:
-h, --help show this help message and exit
-a, --article ARTICLE
Article you want the audio summary to be about
-c, --config CONFIG Path to a TOML file with configuration
-na, --no-audio Don't generate audio output
-nt, --no-transcript Don't generate an audio transcript
-o, --out OUT Path of output MP3 file
-ot, --out-transcript OUT_TRANSCRIPT
Path of output transcript
-q, --question QUESTION
User question that will be included in the summary
-v, --verbose Show debug information
```
## License
MIT © [Hay Kranen](http://www.haykranen.nl)