Save to audio

[GonESI]

Hi, I was wondering if there is any way to save the returned audio to any audio format.

Thanks in advance (the voice is superb).

[desbma]

Hi,
Currently there is no easy way to do this, but you can run google_speech with -v debug, copy the URL, download it, and save it as an MP3 file.

[alphapapa]

It would be really nice to be able to do this. I hacked this together to dump the audio files into /tmp/google_speech:

https://github.com/alphapapa/GoogleSpeech/tree/dump-audio

I had first tried to pull them directly out of the cache database, but they are re-compressed, which means they aren't plain mp3 files. (This is strange, because they are already MP3 files, so compressing them with gzip or bzip2 doesn't gain anything.)

[desbma]

@alphapapa the code in the web_cache.py is generic and I use it in several other projects, but in google_speech the compression is never used.

What is stored in the database is the raw MP3 file, which you can see for example on Linux with:

$ sqlite3 ~/.cache/google_speech/google_speech-cache.sqlite "SELECT data FROM sound_data LIMIT 1;" > /tmp/a.mp3
$ file /tmp/a.mp3 
/tmp/a.mp3: MPEG ADTS, layer III, v2,  32 kbps, 16 kHz, Monaural

[alphapapa]

@desbma Oops, I didn't read the code closely enough, I just hacked something together. :) I had tried copying the data out of the database with a SQLite database editor, but I guess it couldn't handle binary data. Thanks for your help.

And BTW thanks for continuing to update this project. I used it yesterday for the first time in a while and it didn't work, but I upgraded it with pip and it worked again. Much appreciated!

[emazzotta]

I made a docker container to solve this problem for myself. It's a bit hacky, I'm basically using ffmpeg to record the google_speech audio output.

Maybe it could be interesting for someone else: https://github.com/emazzotta/tts2audio