[linux-audio-user] Re: linux-audio-user Digest, Vol 7, Issue 35

philicorda philicorda at ntlworld.com
Sat Apr 17 14:47:51 EDT 2004


>Date: Fri, 16 Apr 2004 09:43:00 -0700 (PDT)
>From: R Parker <rtp405 at yahoo.com>
>Subject: [linux-audio-user] sndfile-info and lots of questions
>To: linux-audio-user at music.columbia.edu
>Message-ID: <20040416164300.90068.qmail at web20916.mail.yahoo.com>
>Content-Type: text/plain; charset=us-ascii
>
>Hi,
>
>As I've gone through my most challanging mastering
>project, I've developed more questions than answers. 
>  
>
Hope it all turned out right in the end!
Only just caught up with the last email. I reackon Jamin could easily 
sort out the occasional uncontrollable kick drum. Perhaps in a two pass 
process, once to sort out that, then another pass for more general 
mastering.

>There's been reference, on this list, to documentation
>that explains file formats-- I don't recall the
>document title or where to find it. Ultimately, I need
>a resouce that explains things like; the number of
>available samples for the different bit depths (16bit
>range from -ABC to +XYZ, 24bit...), DC offset is, peak
>amplitude is, RMS is, etc.
>
>I need to know if samples are syncronous with decibel
>level, is maximum samples equal to 0db?
>  
>
Yes. Though 0dbfs is perhaps a better way to put it.

Samples in 16bit= 2^16 Samples = -96dbfs range.
Samples in 24bit= 2^24 Samples = -144dbfs range.
Samples in 32bit float = I'm not sure if you count the Mantissa or not = 
Shockingly massive dynamic range. (Almost exactly :) ).

>Of course another challange is tools like sndfile-info
>assume that a file exists. This is not always the case
>and in my situation it's almost never true. I return
>JAMin output to an Ardour return bus and don't produce
>a file until the return bus is exported. Printing a
>track to the file system and then analyzing it is no
>way to save time.
>  
>

I guess the only way some of those figures can be found is for the file 
to be printed. Ie, you can only tell what the peak sample is if you have 
compared it to every other sample in the file. To do it in real time, 
you would have to have 'Peak Sample so far', 'RMS so far' etc, and reset 
the figures manually when you started playback again. Not to say such a 
facility would not be useful. It would be nice if Jamin had a statistics 
page that showed you how many overs there had been, how many of them 
were consecutive etc.

There is a new program called'Jackbitscope'
http://www.ecs.soton.ac.uk/~njl98r/code/ladspa/bitscope/
this has displays for highest and lowest sample so far. (And lots of 
other useful looking things I don't really understand. )

>Anyway, I appreciate all the responses to my past 
>questions and am hopeful that someone can look at the
>current mumbo jumbo and prescribe some effective
>medications; coffee, sleep, black bear gallbladders,
>urls to useful documents, etc.
>  
>
Ground lion's teeth are effective in 90% of these cases in my 
experience. If not.....

"0dbfs+ levels in digital mastering"
http://www.studio-systems.com/audiofeatures/SepOct2002/Audio%20Article/0DBFS+%20LEVELS%20IN%20DIGITAL%20MASTERING/78.htm
It's a couple of years old but still interesting, in that it covers what 
can happen when very hot digital levels hit the analog world.

Also, the rec.audio.pro faq is chock full of useful stuff...
http://www.faqs.org/faqs/AudioFAQ/pro-audio-faq/




More information about the Linux-audio-user mailing list