Hi, so I wanted to change the response mycroft makes when activated by the wakeword. I saw this post which was very helpful:
I didn’t want to reply there because I don’t know how acceptable it is to revive dead posts or not, so I’m making a new one (lmk if that’s what I should have done or no)
Anyways, so I’m trying to use this sound file as my wakeword response:
So here were my steps:
cd /home/pi/mycroft-core/mycroft/res/snd/
wget https://i1.theportalwiki.net/img/b/bb/GLaDOS_taunt_small_wave02.wav
mkdir old_files
mv start_listening.wav old_files/
mediainfo GLaDOS_taunt_small_wave02.wav
This was the mediainfo output:
General
Complete name : GLaDOS_taunt_small_wave02.wav
Format : Wave
File size : 250 KiB
Duration : 2 s 899 ms
Overall bit rate mode : Constant
Overall bit rate : 706 kb/s
Writing application : Lavf54.59.106
Audio
Format : PCM
Format settings : Little / Signed
Codec ID : 1
Duration : 2 s 899 ms
Bit rate mode : Constant
Bit rate : 705.6 kb/s
Channel(s) : 1 channel
Sampling rate : 44.1 kHz
Bit depth : 16 bits
Stream size : 250 KiB (100%)
Format looks correct except only 1 channel, so not stereo. So I converted to stereo:
ffmpeg -i GLaDOS_taunt_small_wave02.wav -ac 2 GLaDOS_taunt_small_wave02_stereo.wav
mv GLaDOS_taunt_small_wave02.wav old_files/
mycroft-config reload
mycroft-start all restart
When I do all this, I just get no wakeword response. Mycroft still works fine, just no sound when it’s woken up. When I move the old files back to where they were and keep all my changes in the old_files/ directory, I’m able to get the original wakeword response back. So I feel like it’s the format of the file? But it’s wav-format, 16-bit, stereo, 44100 Hz sampling rate. Any ideas? I didn’t have something to play the .wav natively but I did convert to mp3 and was able to play it through my pi’s speakers using mpg123, so it doesn’t seem corrupted or anything.
Edit: Almost forgot, this is using Picroft 21.2.1 on a Raspberry pi 4 Model B. I did also try checking the mediainfo of the original start_listening.wav, which had a 48khz sampling rate, so I tried changing that with ffmpeg as well with the same results.
Thanks!
Wilson