Audio capture problem

[bchinnari]

Hi all

We have configured kaldi gstreamer as per the instructions. We are using kaljurand client package (as mentioned in your readme.md). After this whole setup, when we are trying to record the audio...server/worker is getting request from the client but the raw data which is stored in the tmp folder is empty (0 bytes) and the opened session is idle forever . Kindly let me know where it can go wrong.

Thanks Bhargav

[alumae]

Are you using HTTPS? Newer Chrome versions don't allow recording audio if the page is served over HTTP.

[bchinnari]

We are using firefox.

var SERVER = "ws://localhost:8888/client/ws/speech"; var SERVER_STATUS = "ws://localhost:8888/client/ws/status"; var REFERENCE_HANDLER = "ws://localhost:8888/client/dynamic/reference";

We are using above details in dictate.js and demo.js

[alumae]

Is there anything printed to the Firefox Javascript console? Try alos with Chrome and check the JS console.

[bchinnari]

Using //@ to indicate sourceMappingURL pragmas is deprecated. Use //# instead[Learn More] jquery-1.10.2.min.js:1 navigator.mozGetUserMedia has been replaced by navigator.mediaDevices.getUserMedia dictate.js:113:4 SecurityError: Failed to load worker script at "/home/aai/Downloads/Kaljurand-dictate.js-7efd3b1/lib/recorderWorker.js" (unknown) Use of getPreventDefault() is deprecated. Use defaultPrevented instead. demo.html

This is the console log we are getting.

[alumae]

Does it work with Chrome?

[bchinnari]

No. Same error. We checked the path /home/aai/Downloads/Kaljurand-dictate.js-7efd3b1/lib/recorderWorker.js also. This is the correct path.

[bchinnari]

Hi Tanel,

Finally, it worked after after launching chrome as follows "google-chrome --allow-file-access-from-files" and then opening demo.html cleared the error.

It is a very cool project you developed and working with minimal latency.
Thank you so much for patiently answering the queries. Also, I would like to know more how this is working internally. I would appreciate if you throw some light on this.

[bchinnari]

Hi Tanel,

I have a small question. In the current configuration, we can pre-define number of workers we want to run. Is it possible to launch and close a worker when a new status listener comes and leaves?

[alumae]

No, there is no such possibility.

[bchinnari]

With small hack, we are able to launch the worker when status listener arrives. But we are not able close the worker when listener leaves. Could you help in this.

The hack is one statement

class StatusSocketHandler(tornado.websocket.WebSocketHandler):

def check_origin(self, origin):
    return True
def open(self):
    logging.info("New status listener")
    **os.system("supervisord -c /home/aai/supervisord_1worker.conf")**
    self.application.status_listeners.add(self)
    self.application.send_status_update_single(self)

[alumae]

Sorry, you are on your own with this.

[bchinnari]

I would like to know if it is possible at all