[LAU] question about normalize-audio

shane richards shanerich at email.com
Fri Jul 22 17:08:25 UTC 2011



 
  ----- Original Message -----
  From: Paul Davis
  Sent: 07/22/11 07:18 AM
  To: shane richards
  Subject: Re: [LAU] question about normalize-audio
 
   
>> * normalize the RMS volume of a group of samples

> in general, "normalization" tends to refer to operations on the actual
> sample values, not an average computed number like RMS.

>> * max value of the loudest peak in the group doesn't exceed  xxx db

> if it doesn't do that, its not really normalization.

>> * no limiting/clipping/compression

> normalization would never involved any of these 3 operations. if it
> did, it would be called limiting or clipping or compression.

I'm commenting specifically on the "normalize-audio" program (I'm not aware of anything else), which computes either by peak or RMS, but when computing by RMS the fact that a peak may go over 0db is compensated for by limiting/clipping, instead of changing the gain value for all samples to keep everything under a given limit. It seems you can have one or the other, but not both.

In short, I want the "perceived loudness" values levelled without losing any peaks, and without denormals being introduced.

> are you asking about how to do this as a user (i.e. "what tools are
> available to let me do this?") or as a programmer/developer/tinkerer
> (i.e "how would you go about doing this?")

As a user. 500+ drum samples to be converted into Hydrogen kits. So something scriptable if you know of any.



Shane Richards
Producer, Composer, Multi-instrumentalist
Josh Music
shanerich at email.com
www.josh.com.co
www.shanerichardsmusic.com


More information about the Linux-audio-user mailing list