Add device option

[LinusU]

Same as #91 and #104 but with some improvements:

• handles ao->device lifecycle
• follows coding style
• with a minimal test

I haven't been able to test this myself yet since I'm only on macOS...

ping @meldron, @MexXxo, @T-vK, @rhclayto; would you mind testing if this works? ❤️

[meldron]

@LinusU works for me.

Linux x86_64 node v8.3.0

May be you should add the new option to the README.md.

Great work!

Edit: you could also include how to address different sound cards:

hw1:0: sound card 1 (second card), subdevice 0 (default device)

cat /proc/asound/cards for sound card information in linux

[T-vK]

I only have Linux, so I could only test the same that meldron has already tested.
I agree this should be documented in the readme. A link to this would be helpful for Linux users. For SoX the format for the device name is probably very different. So having that documented would be nice, too.

[cache91]

This is throwing errors and breaking when i try this branch out

[LinusU]

@cache91 what errors are you getting?

[cache91]

It basically shows it cannot open device

[meldron]

@cache91 can you give us more information?

Which os, which node version, what device... Stuff like this

[cache91]

[../deps/mpg123/src/output/alsa.c:103] error: initialize_device(): rate 44100 not available, using 48000
{ audioFormat: 1,
  endianness: 'LE',
  channels: 1,
  sampleRate: 16000,
  byteRate: 32000,
  blockAlign: 2,
  bitDepth: 16,
  signed: true }
events.js:182
      throw er; // Unhandled 'error' event
      ^

Error: specified PCM format is not supported by "alsa" backend
    at Speaker._open (/home/pi/speaker-test/node_modules/speaker/index.js:91:13)
    at Speaker._write (/home/pi/speaker-test/node_modules/speaker/index.js:175:23)
    at doWrite (/home/pi/speaker-test/node_modules/readable-stream/lib/_stream_writable.js:406:64)
    at writeOrBuffer (/home/pi/speaker-test/node_modules/readable-stream/lib/_stream_writable.js:395:5)
    at Speaker.Writable.write (/home/pi/speaker-test/node_modules/readable-stream/lib/_stream_writable.js:322:11)
    at Reader.ondata (/home/pi/node_modules/readable-stream/lib/_stream_readable.js:572:20)
    at emitOne (events.js:115:13)
    at Reader.emit (events.js:210:7)
    at readableAddChunk (/home/pi/node_modules/readable-stream/lib/_stream_readable.js:195:16)
    at Reader.Readable.push (/home/pi/node_modules/readable-stream/lib/_stream_readable.js:162:10)

This is the error that I am getting. The first console.log is the format that I am passing. I was not getting this problem with the master branch

[cache91]

       reader.on('format', function (format) {
console.log(format);
 speaker1 = new Speaker({
device:'plughw0:0',
audioFormat: 1,
  endianness: 'LE',
  channels: 1,
  sampleRate: 16000,
  byteRate: 32000,
  blockAlign: 2,
  bitDepth: 16,
  signed: true});
  reader.pipe( speaker1);
                            });

This is my code. Using raspbian OS on node 8.7.0

[T-vK]

Sounds like a problem on your end. What happens if you use the same options with aplay on the command line?

[cache91]

@T-vK : the file plays on the command line after i save it. It seems so wierd. Also the fact that with the same hardware and setup , I am able to play using the package in master branch

[T-vK]

@cache91 Hm, maybe it is a problem with this PR then. I have successfully used this project on raspbian before. I used the version that I forked some time ago. https://github.com/T-vK/node-speaker/tree/select-audio-device (the select-audio-device branch)

This code has worked for me on Raspbian: https://github.com/T-vK/tvoip/blob/master/audio-test.js It simply pipes the mic input to the speaker. I have used a Pi Zero W and thus I had to use a USB sound card btw.