How to read the output of NFA (Nemo Forced Aligner Tool) ?

[Ca-ressemble-a-du-fake]

Hi,

I tried NFA to align a sentence with audio thanks to citrinet fr version. Yet I don't fully understand the output :

<utt_id> 1 <start time in samples> <duration in samples> <text, ie token/word/segment>

Duration in samples is given as a decimal value like 19.52. I would expect it to be an integer representing the number of samples at a sampling rate of 16kHz. But this is not the case.

So how to convert this duration in samples to a duration in seconds ?

Thanks in advance for your answer,

[titu1994]

According to @erastorgueva-nv
The float values represents the time in seconds.

[Ca-ressemble-a-du-fake]

Thanks I'll take note of this.