1
$\begingroup$

I seek a scaled audio file format for my dsp output.

I'd like to produce "scaled wav files", when I've done my pre processing. Eg. if the amplitude is 15.5 (of whatever unit), you should be able to open a standard wav/whatever file and see "15.5".

If you produce a normal 16 bit wav file, you have to tell the user, to scale all values with value xx.xx. This would be ok, if the scale factor could be included in the file, in a standardized fashion. But so far I haven't been able to find any "scale factor" field in the wav format?

Another way could be to produce a 32 bit float wav file. This could actually contain the value "15.5". Problem is that wav float values are defined to be in the range [ -1.0 : 1.0 ]. I've tried opening such a file in different freewares like eg. Audacity. But they don't take well to the larger scale. They still only display the [ -1.0 : 1.0 ]. Do you know a program that will display large scale float wav files?

(Btw I do realize that the "scaled audio" is not "audio" anymore.)

Improve this question
$\endgroup$

3 Answers 3

Reset to default
1
$\begingroup$

ID3v2 tags include RVAD (Relative Volume ADjustement), and two of the RVAD values are "Peak volume right" and "Peak volume left".

The problem is not fixed point versus floating point though. The two are pretty similar. You can treat the 16 bit fixed point as a integer (Q15.0), or as a x/32768 fraction (Q0.15) so the range is -1.0, +1.0 or as x/256 (Q7.8, -256,+256 range).

Improve this answer
$\endgroup$
1
  • $\begingroup$ Yes, it seems like the peak value is a valid scale factor. However, according to this: id3.org/id3v2.4.0-changes it's deprecated. The replacement RVA2 also seems useful though. $\endgroup$
    – Illishar
    Commented Jan 3, 2018 at 8:20
1
$\begingroup$

This link implies that .wav files have several standards for storing 32 bit floating point. It looks like foramt 32-bit 16.8 float (type 1 - 32-bit) would do what you want.

As best I can see, SOX (reference) can work (read it at least) with this type of format, so you could look at the source code to see how to generate/read these files. CoolEdit claims to be able read and write this format (see first link.)

Improve this answer
$\endgroup$
3
  • $\begingroup$ It's interesting at least. However, I've looked through the references and the output. It turns out that the CEP 32-bit 16.8 float format is legacy and marked as obsolete. That doesn't matter so much, so I tried saving a file in the format. The 16.8 float format is actually a 32-bit PCM file. (A 32-bit integer.) And there's nothing that indicates (eg. extended subformat) that it should be interpreted as float or anything. In other words, it's a legacy bug. $\endgroup$
    – Illishar
    Commented Oct 14, 2014 at 9:37
  • $\begingroup$ Btw, I'd vote your answer as 'useful', but I cannot seem to vote yet. $\endgroup$
    – Illishar
    Commented Oct 14, 2014 at 9:38
  • $\begingroup$ How is SOX able to recognize it, I wonder ... $\endgroup$
    – Illishar
    Commented Oct 14, 2014 at 9:45
0
$\begingroup$

WAV files in 32-bit WAVE_FORMAT_IEEE_FLOAT can handle any arbitrary value in the range $±3.4×10^{38}$.

I created a tone in Ocenaudio and amplified it to 15.5, saved it and re-opened it, and it works fine. You just have to zoom out from the default −1 to +1 range to see it:

Ocenaudio zoomed out

Actually, I guess it can only zoom out to ~±32000...

Audacity can only zoom out to −2.0 to +2.0.

Sonic Visualiser can zoom out, but clips the signal to ±1.0. :/

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.