Audio transcription with OpenAI Whisper on Raspberry PI 5

Video Description

System folder

The system folder has the script install.sh that will install all required software for the test.

Python folder

The script daemon_audio

python daemon_audio.py

will start a audio recording into a /data folder. The audio stream will be split on 10-second chunks and added to a queue for transcription.

The script daemon_ai

python daemon_ai.py 0

will start a audio transcription processing. It will grab audio files from daemon_audio (if any) and transcibe them into a /data folder. The transcript will be in a data folder in a *_transcription.txt files.

You can change OpenAI whispes models https://github.com/openai/whisper?tab=readme-ov-file#available-models-and-languages :

python daemon_audio.py 0 # tiny.en
python daemon_audio.py 1 # base.en	
python daemon_audio.py 2 # small.en	
python daemon_audio.py 3 # medium.en