The wav file header can contain a chunk specifying maximum amplitude in the soundfile. Without it the software would either assume the max amplitude is 1.0 or have to scan the whole file to find it. You can see a number of things could go wrong here. Try normalizing the amplitude to O.5 max and see what happens.

Sorry, I was thinking of floats. I don't know what happens with 24 bits.

Short of having a specific example to inspect, it is not possible to do 
more than speculate. One possibility is that the file was ill-advisedly 
normalized to digital peak, which would in fact lead to clipping on 
rendering given the signal is non-sinusoidal (and compounded by 
convolution with a high-level source). I have never used VLC; Audacity 
is pretty trustworthy in these things, though it does not interpolate 
graphically between samples to show the underlying analogue waveform the 
way Adobe Audition (nee Cool Edit Pro) does.

It is equally possible that the IR files from that source are inherently 
less than perfect. It may also be necessary (still speculating) that the 
level of the IR needs to be reduced anyway, within the convolver.

Richard Dobson



On 10/06/2016 06:08, Michael Mossey wrote:
> This is not directly csound related, but maybe someone would know what
> to look for. I downloaded some impulse responses from www.samplicity.com
> . When I play the WAV file with VLC, it is
> clearly clipping at the beginning. When I load the impulse response into
> Reaper, the reverb effect is highly distorted--it sounds like it is
> clipped. When I load it into Audacity, it does not look clipped. The
> format is 24 bit, 44.1 KHz. I wondered if VLC can't handle that so I
> exported it as 16 bits from Audacity. It still sounds clipped when
> playing through VLC. Doesn't sound any different in fact.
>
> Any ideas what to look for?
>
> Mike
>
> Csound mailing list Csound@listserv.heanet.ie
> 
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND Send bugs reports to
> https://github.com/csound/csound/issues Discussions of bugs and features
> can be posted here

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

As Richard said above, it could also be just an amplitude related issue:
convolution between two normalized signals can often return to the output
very high amplitude values. In order to theoretically avoid this, the
ReaVerb plug-in that you mentioned, for instance, allows to apply a -18dB
gain to the convolving function, or even to control the level of the
resulting convolved (wet) signal. Being convolution a linear operation, it
doesn't matter if you scale the amplitude of the convolving or the convolved
signal because the result will be the same.



--
View this message in context: http://csound.1045644.n5.nabble.com/Csnd-impulse-response-weirdness-tp5750115p5750155.html
Sent from the Csound - General mailing list archive at Nabble.com.

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here