How to return timestamps in different format?

[c6s0]

How can i return the timestamps in h:m:s format and not in s.ms ?

I guess you should be able to do it somewhere here but i can't find the specified format?

if return_seconds:
    for speech_dict in speeches:
        speech_dict['start'] = round(speech_dict['start'] / sampling_rate, 1)
        speech_dict['end'] = round(speech_dict['end'] / sampling_rate, 1)
elif step > 1:
    for speech_dict in speeches:
        speech_dict['start'] *= step
        speech_dict['end'] *= step

[ryanlong1004]

I'm guessing speech_dict start and end times are epoch timestamps, correct?

[snakers4]

I'm guessing speech_dict start and end times are epoch timestamps, correct?

Not sure what you mean by epochs in this context, but these are indeed seconds.

How can i return the timestamps in h:m:s format and not in s.ms ?
I guess you should be able to do it somewhere here but i can't find the specified format?

There is no format.
These are just seconds with some precision.
Typically less than 50-100ms is not practical in a non-streaming fashion.
If you need another format - you can just override this function and / or replace with yours.

[dementedAngel]

Okay, so I gather that the timestamps are in s.ms format. how many digits are given for the ms? I'm trying to figure out what my data mean!

[snakers4]

silero-vad/utils_vad.py

Line 170 in 262bcb4

return_seconds: bool = False,

[snakers4]

It is either seconds or samples

[dementedAngel]

I figured out that if I divide by 10000, I'm getting about 480 'seconds' in a 5-minute segment. So, to get the real time I need to divide again by 1.6.

[dementedAngel]

So yes, it's samples.

[rjac-ml]

yes the format that returns is in sample rate. that is very useful given that the imput wav are sample rate too. So if you want to chunk/parse that audio wav then is very straight forward with this values