Wow this is most unexpectedly great. I posted in utter despair and now I've got a simple solution that seems to do what I want. Just play two AQs.
Jens, thanks for the idea for graphing. Brian, thanks for steering me away from offline rendering. I was hoping to just get by with the AQ built in leveling to display my waveform. I'm hoping that I won't need to translate into PCM. Though the way I'd do that is with a simple ExtAudioFile write? I've seen some very pretty waveforms on some GIT projects. Some that were based on EZAudio. Also I think it was SoundCloud that wrote up a detailed description of a how to get a highly performant graph. So I'll try those. Thank you! Matt On Wed, Aug 5, 2015 at 7:08 PM, Brian Willoughby <[email protected]> wrote: > > On Aug 5, 2015, at 9:50 AM, Jens Alfke <[email protected]> wrote: > > On Aug 5, 2015, at 9:41 AM, Brian Willoughby <[email protected]> > wrote: > >> The two most common ways to depict audio waveforms are Peak and RMS. I > don't think there's any standard for displaying cube root - although I'm > sure it looks interesting. I'd recommend against inventing new math for > displaying audio waveforms, especially since there are well-established > standards. > > > > I think we’re talking about different things. Peak and RMS describe how > to reduce a time range of a waveform to a single number. What I was > referring to is how to transform that number (i.e. a raw amplitude) into a > volume level that’s meaningful to a human, since our ears’ response to > sound pressure is highly non-linear. > > > > I did some online searching and the two common approaches seem to be > either logarithmic or cube-root. I experimented with both, and the latter > seemed to produce a more intuitively-correct display, to me at least. > (Disclaimer: I could be mis-remembering the formula; it’s been several > years since I looked at my code for it.) > > Ah, yes, the raw samples correspond directly to voltages, and you can > calculate power amplitude from that, but the perceived loudness is not > precisely linear compared to the voltage and power in the circuit. So, > yeah, Peak/RMS are different than what you're talking about. > > It's very tricky to get into this area, though, because frequency content > and frequency distribution significantly affect the relation between what's > happening in the electronics compared to what's perceived in our brains. > The formula for a pure tone at a single frequency is exponential or > logarithmic. But when there are multiple frequencies the translation > becomes more complex. Take a look at the Bark scale and critical bands. Of > course, simpler math can get close without being precise. > > The advantage of Peak and RMS is that they can be exact. The disadvantage > of fancier, perception-based loudness displays is that they're much less > accurate unless you have a lot of information. > > Brian Willoughby > Sound Consulting > >
_______________________________________________ Do not post admin requests to the list. They will be ignored. Coreaudio-api mailing list ([email protected]) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/coreaudio-api/archive%40mail-archive.com This email sent to [email protected]
