I got a pretty slow internet connection. Transcoding the wav recordings to webm (opus or vorbis) or mp3 before sending it to the openAI API might speed up the whole process of STT. Conversion of small audio snippets should not take too long on a recent Raspi and can be done with pydub or ffmpeg itself pretty easily - https://www.geeksforgeeks.org/convert-mp3-to-wav-using-python/.
I got a pretty slow internet connection. Transcoding the wav recordings to webm (opus or vorbis) or mp3 before sending it to the openAI API might speed up the whole process of STT. Conversion of small audio snippets should not take too long on a recent Raspi and can be done with pydub or ffmpeg itself pretty easily - https://www.geeksforgeeks.org/convert-mp3-to-wav-using-python/.