0
0

Hello everyone…

I’m having a problem using FMOD_OUTPUTTYPE_NOSOUND_NRT with getSpectrum.

I’m using the source code of the well-known example pitch detection, and I have used this with an mp3 file not from recording anymore…

the problem is that :
When i let play the song I’m getting a certain result (when not using FMOD_OUTPUTTYPE_NOSOUND_NRT)

and when i use FMOD_OUTPUTTYPE_NOSOUND_NRT, i’m getting another result which is radicly different from the first one, although it is the same mp3 file i’m using…

Should I say also, taht the analyse of the whole file is 15 times faster than without FMOD_OUTPUTTYPE_NOSOUND_NRT. (20 seconds versus 3 minutes when the song is playing)

Any suggestions ?

Thank you.

  • You must to post comments
0
0

Nobody Know why I don’t get the same result ?

  • You must to post comments
0
0

Correct me if I’m wrong but do you get more accurate results?

I think getSpectrumData result depends on the speed in which you call system::update.

  • You must to post comments
0
0

I guess this is one reason, I have thought about it…

But the thing is, even it i the same music I’m playing, a lot of found frequency on the first sampling without FMOD_OUTPUTTYPE_NOSOUND_NRT will never be founded on the sampling with FMOD_OUTPUTTYPE_NOSOUND_NRT

I was prepared to see more accurate results with FMOD_OUTPUTTYPE_NOSOUND_NRT instead of that i’m getting a whole different sampling, that is what I don’t understand…

  • You must to post comments
0
0

[quote="TheSquad":19v5dx8d]I guess this is one reason, I have thought about it…

But the thing is, even it i the same music I’m playing, a lot of found frequency on the first sampling without FMOD_OUTPUTTYPE_NOSOUND_NRT will never be founded on the sampling with FMOD_OUTPUTTYPE_NOSOUND_NRT

I was prepared to see more accurate results with FMOD_OUTPUTTYPE_NOSOUND_NRT instead of that i’m getting a whole different sampling, that is what I don’t understand…[/quote:19v5dx8d]

Other than changing the spectrum window type to blackman harris, I have no other idea to improve your reading…

  • You must to post comments
0
0

[quote="icuurd12b42":jkf84amv]
Other than changing the spectrum window type to blackman harris, I have no other idea to improve your reading…[/quote:jkf84amv]

Blackman Harris is more accurate ?

  • You must to post comments
0
0

[quote="TheSquad":32kbc7js][quote="icuurd12b42":32kbc7js]
Other than changing the spectrum window type to blackman harris, I have no other idea to improve your reading…[/quote:32kbc7js]

Blackman Harris is more accurate ?[/quote:32kbc7js]

Probably not going to fix the problem I think you are having though. This FFT is, from what I have heard, is better at handling data on the edge of uncertainty… for lack of better words.

These are snapshots in time… And I assume they are averaged to hold a good representation of the data acumulated between update() calls.

If my assumption is correct, you are still going to have drastic differences from getting the spectrum 30/60/120/500 times a seconds.

A FMOD expert should take over from here. I hate to discuss things based on assumptions when I know someone out there knows how this feature works exactly.

  • You must to post comments
Showing 6 results
Your Answer

Please first to submit.