Respeaker in Docker Container

[piisku78]

It would be useful if Seeed could provice a sample Docker image for Raspberry which includes ReSpeaker. This could enable usage of ReSpeaker in Cloud IoT Platforms and developers could build intelligent, scalable IoT solutions.

I would like to build Azure Edge module which contains 4 array ReSpeaker. For that I need a container module but I cannot get ReSpeaker installation running inside a Docker container.

---------- in Dockerfile I do like this --------

FROM arm32v7/python:3.7-slim-buster … … RUN apt-get -y install portaudio19-dev python-pyaudio python3-pyaudio RUN git clone https://github.com/respeaker/seeed-voicecard.git RUN chmod 777 seeed-voicecard/install.sh RUN apt-get install -y alsa-utils dkms i2c-tools libasound2-plugins

WORKDIR /app/seeed-voicecard RUN ./install.sh

------ ERROR while building the Docker container ------

[15/15] RUN ./install.sh:

19 0.426 Warning: /boot volume not found …

19 0.430 /boot/overlays not found or not a directory

19 0.463 dtparam not found

19 0.463 You may need to run ./ubuntu-prerequisite.sh

19 0.492 dtoverlay not found

19 0.492 You may need to run ./ubuntu-prerequisite.sh

19 0.494 Errors found, exiting.

executor failed running [/bin/bash -c ./install.sh]: exit code: 1

Why install.sh does not work inside a container? What should I do to get this running?

[HinTak]

Do you expect any kind of hardware device driver to work inside docker?

Disclaimer: I don't work for Seeed Studio.

[piisku78]

well if it is not possible to run install.sh (I dont know what it exactly does) then I can run the install.sh script directly on the device command line while I do initial setup for the device. But then there must be away to get the audio from ReSpeaker to Docker application. How to do it?

[HinTak]

Why do you think it should be possible to run install.sh inside a docker container? A container naturally separates and isolates the user from accessing devices. The isolation is what many people use docker for.

It is possible to pass the audio over a "network" link between a container and its physical host. You search for "network audio" on Google if you want to do that. You still need to put the driver on the host, not on the container.

Disclaimer: I don't work for Seeed Studio.

[piisku78]

I found some old docker images where people have used a custom raspberry pi 3 image as base image. The custom version obviously contains necessary drivers for ReSpeaker mic so a docker image based on that works out of the box and one can access the mic audio data in python code running in a container.

But I will install in my project the drivers on host and then re-route the audio data from hardware to docker container so problem is solved.

[HinTak]

@piisku78 if you know of such a docker image, you should post the url. You should be able to run apt update; apt upgrade and save, to update the docker image to current!