Increase maximum volume for TTS output

[ClemensGruber]

We did some tests in loud environment and noticed a too low volume with the Anker PowerConf S330 USB conference system, even with volume control of the speaker on max position. While playing a seconde sound file we noticed that the parallel played second file was much louder than the voice of GPTree and so global system / speaker volume seems not to be the limitation. The speaker can output loud sound!

1. Check if the used software player could be the problem:

• For the second file we used ffplay as software player.
• For the first (GPTree) file we used os.system(f"afplay {filename_output}")
• Check if there is a volume parameter in ffplay vs. afplay what could cause the difference

2. Check if the encoded file has to less volume

• Perhaps the TTS created voice is too quiet and we have to increase the volume on this side.
• There are some indicators that this is possible, see the Google TTS API documentation https://cloud.google.com/text-to-speech

• Volume gain control: Increase the volume of the output by up to 16db or decrease the volume up to -96db.

[ClemensGruber]

1. Check if the used software player could be the problem:

   For the second file we used ffplay as software player. For the first (GPTree) file we used os.system(f"afplay {filename_output}") Check if there is a volume parameter in ffplay vs. afplay what could cause the difference

Inspect both files and analyze it on PC

• We have started GPTree and initiated a conversation so that a response audio file was generated from the Google Cloud service.
• You can find the file in folder /audio file output.mp3
• We compared this file with pixabay.com_are-you-sleeping-145026 what is played much louder on the RasPi.
• Hearing the two files on PC leads to the same ovservation: GPTree is less loud than the pixabay file, But the pixabay file sounds overdriven and the recording level seems not correct.
• Loading both files in Audacity and inspecing waveform shows the difference:
• 

So different volume levels are caused by the recording, not by the (software) player we used. The pixabay file sound overdriven (much more on the PC than on the Anker speaker), this is "too much", but we have room for improvement on the Google TTS service file and will try to improve volume range to get a louder sound!