Aditya Posted November 4, 2019 Posted November 4, 2019 Armbianmonitor: http://ix.io/20OD Hi ! I am using Armbian_5.65_Orangepilite_Debian_stretch_next_4.14.78armbian deb stretch image on orange pi lite. Its working great for the most part, but it misses the starting portion of the audio. I have installed snips voice assistant on it and for some reason the speaker does not produce the confirmation sound output on hotword detection. But plays rest of the sound outputs after that for another 5 seconds or so. Missing the confirmation sound from a smart speaker is really annoying. Is that a bug? Is there a possible workaround? Is any way this can be solved properly? Please point me in the right direction
Werner Posted November 5, 2019 Posted November 5, 2019 Your Armbian is quite old. Have you tried to use a more recent image? This issue may have been fixed in the meantime.
Aditya Posted November 5, 2019 Author Posted November 5, 2019 42 minutes ago, Werner said: Your Armbian is quite old. Have you tried to use a more recent image? This issue may have been fixed in the meantime. Okay I would try that and revert back. But the problem is I am using the board solely to run Snips AI and Snips has already mentioned on their website that they do not support debian buster yet. Only Deb Stretch is supported.
Werner Posted November 5, 2019 Posted November 5, 2019 I think for a simple test if a more recent image fixes the issue the running OS should not matter. Everyone should have spare sd cards for such situations Besides that Armbian has not been abandoned Stretch. Such images can still be built using the build script: https://github.com/armbian/build
guidol Posted November 5, 2019 Posted November 5, 2019 1 hour ago, Werner said: Your Armbian is quite old. Have you tried to use a more recent image? This issue may have been fixed in the meantime. I dont think its fixed, because I also got this on Armbian Linux 5.3.8-sunxi package bsp-kernel[5.99.191102] u-boot[5.91] dtb[5.99.191102] firmware[5.98] config[5.98] on a OPi Zero. Also got this "problem" since over 2 years: Since then I use my old workaround from this old thread. 1
Werner Posted November 5, 2019 Posted November 5, 2019 Thanks for the follow-up. It was a pure guess and I think it makes sense to support the most recent (stable/next) version of Armbian only to avoid talking over and over about may or may not already fixed issues
Aditya Posted November 5, 2019 Author Posted November 5, 2019 2 hours ago, Werner said: I think for a simple test if a more recent image fixes the issue the running OS should not matter. Everyone should have spare sd cards for such situations Besides that Armbian has not been abandoned Stretch. Such images can still be built using the build script: https://github.com/armbian/build So, today I tried this one also- Armbian_5.83_Orangepilite_Debian_stretch_next_4.19.38. But the same result. Later I tried the latest buster image- Armbian_5.91_Orangepilite_Debian_buster_next_4.19.59. But could not successfully run the software I wanted on that and hence could not test it. Maybe I would give it another try some day
Aditya Posted November 5, 2019 Author Posted November 5, 2019 2 hours ago, guidol said: I dont think its fixed, because I also got this on Armbian Linux 5.3.8-sunxi package bsp-kernel[5.99.191102] u-boot[5.91] dtb[5.99.191102] firmware[5.98] config[5.98] on a OPi Zero. Also got this "problem" since over 2 years: Since then I use my old workaround from this old thread. Sorry, dn't know how to use that workaround for picotts used by snips Can you guide me a bit please.
guidol Posted November 5, 2019 Posted November 5, 2019 2 hours ago, Aditya said: Sorry, dn't know how to use that workaround for picotts used by snips Can you guide me a bit please. find the confirmation sound-file for the hotword detection and add 3-5 seconds of silence to the beginning of the sound-file (maybe via audacity) or upload the file and I do try to add the silence-seconds via audacity for youhttps://forum.snips.ai/t/change-wake-word-response/2419/3 Quote koan Apr 25 The sound feedback option in /etc/snips.toml is both for the feedback after the wake word is detected and after you have spoken your command. You can change the wav files in /usr/share/snips/assistant/custom_dialogue/sound/, but they will be overwritten after each update of the assistant.
Aditya Posted November 5, 2019 Author Posted November 5, 2019 9 minutes ago, guidol said: find the confirmation sound-file for the hotword detection and add 3-5 seconds of silence to the beginning of the sound-file (maybe via audacity) or upload the file and I do try to add the silence-seconds via audacity for youhttps://forum.snips.ai/t/change-wake-word-response/2419/3 Thanks a lot. I'll try that. If I need help in that, I'll come back.
Recommended Posts