Overs... working correctly?

[Naturally Digital]

Hi all,

I'm seeing some funny behavior here...

While playing a stereo wav in SS, the Mixtreme and digicheck meters are registering overs while the SS meters are not. I tried patching the Sonoris Meter and it isn't registering the overs either, even when set to overcount=1.

When I look at the waveform in Soundfile view, it is flat-topped, but on the negative side of the waveform only. In one section there were 10-12 consecutive clips.

This was an empty EDL, no plugins patched and no processing. All faders at unity.

[PieterS]

Hi David,

the Sonoris meter registers an over when the signal equals or exceeds the max. or min. level. This is absolutely sample accurate, in other words, it shouldn't mis anything.
Check the exact level in the soundfile view please.

[Naturally Digital]

Hi Pieter,

In one of the sections, this is what I found:

4 consecutive samples, values (in order): -32767 (FFFF8001), -32741, -32624, -32767

The Mixtreme meters register overs on this section.

[SoundSuite]

David,

Is the wav file by chance normalized to 100%?
(I think? I remember? a thread similar? to this a while ago that this was the culprit of 'x' reporting it overed and SAW/plugs not reporting it as an over.)

It's been a loooong night/day though, I'm about cooked honestly.

[Naturally Digital]

David,

Is the wav file by chance normalized to 100%?
(I think? I remember? a thread similar? to this a while ago that this was the culprit of 'x' reporting it overed and SAW/plugs not reporting it as an over.)

It's been a loooong night/day though, I'm about cooked honestly.

Hi Jon,

Well, I'm disappointed to say that this happens to be Alison Krauss' New Favorite CD. (one of *my* favorites ) I'm in the process of ripping my CD collection to harddrive and setting up playlists in SAWStudio. I ripped it using Wavelab and didn't convert/process it in any way. I've got about 30 of my CD's done so far.

Now that I see what's going on with the waveforms, I understand why there are a few spots on this CD that sound a little 'edgy'. These spots correlate well with the 'overs'.

At first I wondered about some kind of funny copy protection thing but now that I've seen the waveforms, I'm not convinced.

[AudioAstronomer]

Hi Pieter,

In one of the sections, this is what I found:

4 consecutive samples, values (in order): -32767 (FFFF8001), -32741, -32624, -32767

The Mixtreme meters register overs on this section.

Those arent technically full over's. ... probabaly why.

You can keep in mind that even with samples 6db below 0dbFS, overs can occur in DA conversion so.. .its quite an interesting science to determing what really is an over.

[PieterS]

David,

can you send me the sample? I'll look into it.

[Naturally Digital]

David,

can you send me the sample? I'll look into it.

Done.

FWIW, Wavelab doesn't register this as an over (clip) either.

[Naturally Digital]

Those arent technically full over's. ... probabaly why.

Well, Sydec and RME would beg to differ... as would I

LOL! Anyway... for the sake of conversation, what *is* the max value on the negative side of the scale -32768 or -32767? The soundfile view indicates the -32767 as 100% and 0.000000 db.

Robert! Overs at -6dbFS during DA conversion??? WHAT in the name of digital audio science are you talking about?

[AudioAstronomer]

Well, Sydec and RME would beg to differ... as would I

LOL! Anyway... for the sake of conversation, what *is* the max value on the negative side of the scale -32768 or -32767? The soundfile view indicates the -32767 as 100% and 0.000000 db.

Robert! Overs at -6dbFS during DA conversion??? WHAT in the name of digital audio science are you talking about?

audio is not drawn from sample to sample. There are crests between samples that in rare cases can be up to 6db above the highest sample.

Most cheaper (and many prosumer) converters dont deal will with that occurance.

Either way, +/- 32767 is the max for a 16bit file Im fairly sure.. but Ive never known a single sample to be construed as clipping. I missed the point where you said you set the sonoris meter to 1 sample though... it should trigger in that instance eh?