You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hey, i literally tried everything. @jmdobry could you look?
The response is always empty, when using the nodejs-docs-samples/speech/recognize.js with parameter listen
The examples with use of reading from the file works fine.
I had to change line 16 to: recordProgram: 'sox'
in file
\node-record-lpcm16\index.js
because the program uses "rec" by default, not "sox", and with "rec" there is an error, because there is not such file
for troubleshooting, i edited line 226 in file recognize.js to: .on('data', (data) => {process.stdout.write(data.results); console.log(data);}
and the result is:
previously did "install" command as described in the readme
win 8.1 pro
sox 14.4.2 from official site https://sourceforge.net/projects/sox/files/sox/14.4.2/ && path added to environmental "path" ( "sox" typed in cmd.exe works fine)
Node.js v 7.5.0
Moreover, depending on the signal level, the process ends after 0.5 - 5 seconds, but no matter how good audio is (i am using kinda "loopback", so the audio from the american native speakers recorded in good quality gets directly to mic input of the computer)
The audio was thoroughly tested, and the same audio when streamed from file, as a file, not realtime streaming, is recognised with 95% efficiency (about 5% words misunderstood)
tried version 0.1.4 of the rec library (as suggested in one of threads by @jmdobry is guess) (also with change: line 26 to var cmd = 'sox'; from var cmd = 'rec'; same result - not working
Activity
JJablon commentedon Feb 17, 2017
Moreover, depending on the signal level, the process ends after 0.5 - 5 seconds, but no matter how good audio is (i am using kinda "loopback", so the audio from the american native speakers recorded in good quality gets directly to mic input of the computer)
The audio was thoroughly tested, and the same audio when streamed from file, as a file, not realtime streaming, is recognised with 95% efficiency (about 5% words misunderstood)
JJablon commentedon Feb 17, 2017
tried version 0.1.4 of the rec library (as suggested in one of threads by @jmdobry is guess) (also with change: line 26 to
var cmd = 'sox';fromvar cmd = 'rec';same result - not working
JJablon commentedon Feb 17, 2017
gillesdemey/node-record-lpcm16#8
this probably helped (so far), issue to be closed
chore: release 3.2.1 (#316)
chore: release 3.2.1 (#316)
chore: update lint rules and synth (#316)
chore: update lint rules and synth (#316)
chore: update proto comments and fix tests (#316)
chore: update proto comments and fix tests (#316)
migrate code from googleapis/nodejs-speech (#2857)
docs: fix docs 404 in samples/docs (#316)
docs: fix docs 404 in samples/docs (#316)
docs: fix docs 404 in samples/docs (#316)