Removing Noise with AI - RM Noise

I can´t send the records, the system always tell me "The recording is shorter than 30 minutes and has been rejected", and it´s not true, the record has 30 and more minutes. I try it for five times, three are rejeted and two loss.

I'm sorry you are having trouble. I too only see 3 [short, rejected] files.

I restarted the filtering, but no additional files were found. (One of the servers is offline, and I don't know if that is part of the issue.)
Another possible bug is that the server doesn't do enough to ensure the filenames are unique during the same session, so I will add this to my todo list.


Please try again, and use different descriptions for each recording or close the RM Noise client between recordings.

I have deleted one file, and i´ve uploaded another one with other name, and the same thing has happened "the file has been rejeted because is shorter then 30 minutes".

I examined the logs. Here are my observations:

a) indeed 30 minutes expired between starting and stopping the recording

b) however, only 11.8 minutes of data packets were received, which matches the length of the recording file

c) no significant packet loss was observed for the first 11 minutes

d) but, the logging of delay, loss, etc stopped also at 11 minutes, while other logging continued normally

e) nothing I see in the log explains why the delay/loss messages stopped. I specifically looked for: 1) stopping the stream and 2) standby mode

f) on 6/2 other users using rmnoise3 have file lengths that exactly match the recording time over the same time period


I cannot explain this behavior. Can you recall if you changed any setting, 11.8 minutes after starting the recording ?

I have put the PC at maximum performance, I have not let Windows go into sleep mode, I have made another recording and it is in unchecked recordings. A good idea would be to have two options: one, make the recording on the PC and two, manually upload it to the server, but the program does not have that possibility. You could make a version with that option and that way you wouldn't waste so much time trying to upload a file. Greetings.

I absolutely saw logging after the 11.8 minute mark, but there was no logging related to the audio stream - as though it was stopped [by windows].

Like you, I'm also suspecting a mode of windows sleep or alternatively the screen lock.


If you have time to experiment, I would appreciate your help with troubleshooting:

1) Is the problem related to a) performance mode b) 'no sleep' or c) automatic locking the screen (manually done with WINDOWS_KEY + L)

2) If the answer to #1 is C, and if you are using "windows sound mapper" for audio input or output, does it help to explicitly choose the correct sound devices in RM Noise -> OPTIONS -> INPUT DEVICE and OUTPUT DEVICE

The problem should be that Windows went into partial sleep mode, you have to:
- Enter edit energy
- Change advanced power settings
- Choose the plan
- USB configuration
- USB Selective Suspend Setting: Disabled
It will also prevent the screen from turning off and the power plan avoid into sleep mode.
Now I have the files in submitted recordings, what do I do now? How do I pass them to recordings accepted and pending training?
Regards

Now I have the files in submitted recordings, what do I do now? How do I pass them to recordings accepted and pending training?

I perform this step manually. I ensure recordings do not contain speech, tones, etc, and then I accept them.

Afterwards, I periodically train the neural net which can take 8-10 days for each filter type.