Improved performance

[Amjad50]

Methods to improve performance

Trying to improve further

[Amjad50]

I think this is a good improvement, we can do better, but we would need to change hound, as the slowness right now is from hound reading samples.

Encode:

Decode:

As you can see, most of the time is spent on reading the samples, for encoding, its 50/50 between reading from hound and writing to the file, so I think its okay.

But we can make this very fast by using a simple wav writer, and we can do optimization when reading/writing

[TheAwiteb]

Thank you @Amjad50 for this useful improvement

[TheAwiteb]

But we can make this very fast by using a simple wav writer, and we can do optimization when reading/writing

I'll check it

[Amjad50]

Another interesting detail/cheat, we can do this extremely easily and make it insanely fast.

the wav file stores the PCM samples directly as bytes, without any compression, so it means the file content are actually stored in bytes, and this is true, as u can see here.

original file

wav

So we can add just a header for wav and just copy the whole file, which should be fast

[TheAwiteb]

So we can add just a header for wav and just copy the whole file, which should be fast

Wow, is that really possible!! How can we get the header? Is it the same in all files or does it change?

[Amjad50]

Hound uses this library: https://docs.rs/wav/latest/wav/index.html

this library allows u to write the header, so yeah, would be interesting to try.

The header looks like it has basic info, like filesize, sample rate, format, etc....