DIGI-More observations on click/pop problems
 

i am posting this to help Digi figure out these problems and to see if others are able to confirm the same observations that I have made below.

I hadn't had the end of region click problem on 6.4cs7 thru cs9 but now am getting it after updating to 6.7. Strange. Now I have 2 annoying click problems and here are my observations after doing DOZENS of tests:

(1) End of Region Click: (ie, the one that occurs only when recording with ADC on, even if played back later w/o ADC. occurs in regular non-destructive mode, whether a selection is made or recording from the start point and pressing stop to end):

Observation: if I leave approximately 5 seconds or more of silence before pressing Stop, the click does not occur. In this case, if I stop playing my guitar and silence the strings, then wait 5 or more seconds before pressing Stop, there is no end of region click when recording with ADC. I thought this was really strange and doubted it, so I tried it at least 20 times, stopping at various points in time, and it happened every time. Can anyone else confirm whether this happens for them or not?

(2) Clicks when stopping playback if a track is record-enabled (known issue 38454 in PT6.7 Read Me):

Observation: The click only occurs if stopping playback after the end of the region on the track which is record-enabled. Doesn't happen if stopping within the region. Also, it seems that if you press stop outside of the end-of-region but at least 10 seconds or so (variable) after the last audible audio on the track, the click does not occur. This wasn't 100% reproducible for me, but it was about 90% reproducible. So not as certain about the meaning of this observation.


Specs:
PT6.7
HD3
192 i/o
G5 Dual 2GHz
OSX 10.3.4

Re: DIGI-More observations on click/pop problems
 

Hi,

Thanks for the input. Are you clocking your system to any external devices? Do you have any thing connected via AES/EBU, TDIF, or Optical? What kind of audio drives are you using. We appreciate any details that you can give us.
Thanks again,

Digidesign Tech Support

Re: DIGI-More observations on click/pop problems
 

Not clocking to any external devices. No digital connections (no AES/EBU, TDIF or Optical)... only analog. Using Seagate Serial ATA 160GB 7200RPM 8MB cache (second internal drive dedicated to Pro Tools audio only). I get this error even recording onto a single track with minimal playback track count (haven't tried it with zero tracks playing back).

Re: DIGI-More observations on click/pop problems
 

Hi,

Thanks for the added information we are still trying to get it to happen in hoes if you run across anything else you think will help please let us know,
Thanks,
Digidesign Tech Support

Re: DIGI-More observations on click/pop problems
 

ok, I have some more info that may help with the "end of region click with ADC on" problem:

I now am able to reproduce this 100% of the time, but to do it, things have to be timed in a certain way. This is what happens:

First of all, it isn't really a "click" at the end of the region, it is a very short snippet of audio from almost exactly 3 seconds prior to the end of the recording that appears at the end of the region (others have mentioned that they hear bursts of a vocal at the end... this is why). Here's what happens:

(all numbers below were done in 44.1khz/24 bit sessions, WAV format, enforce mac/pc compatibility)

- The snippet of audio at the end of the region is equal to the length (within a few samples) of the maximum delay compensation (ie, the dly+cmp for the track for which dly+cmp is the maximum). So use plugs that create a long delay to see a bigger snippet.

- The origin of the snippet is the audio that starts at a position exactly (within a couple samples) 131,072 samples before the beginning of the snippet (which in turn is starts a number of samples equal to the max dly+cmp before the end of the region/recording). So about 3 seconds prior to the end of the recording.

So if the recording is shorter than 3 seconds, or there is more than 3 seconds of silence at the end, I don't get any "click"... or at least it is not audible.

Digi, here are some numbers you can examine:
(using plug-ins that created longer delay comp made the snippets longer and made it easier to identify from where they came)

(all numbers in samples)
Recording1:
max dly+cmp = 2682
region length = 252804
snippet length = 2698
starting location of snippet audio source:
from start of region = 119034
before beginning of snippet = 252804 - 119034 - 2698 = 131072

Recording2:
max dly+cmp = 127
region length = 161045
snippet length = 125
starting location of snippet audio source:
from start of region = 29848
before beginning of snippet = 161045 - 29848 - 125 = 131072

Recording3:
max dly+cmp = 2708
region length = 165494
snippet length = 2707
starting location of snippet audio source:
from start of region = 31715
before beginning of snippet = 165494 - 31715 - 2707 = 131072

Recording4:
max dly+cmp = 2692
region length = 175100
snippet length = 2697
starting location of snippet audio source:
from start of region = 41331
before beginning of snippet = 175100 - 41331 - 2697 = 131072

Recording5:
max dly+cmp = 5364 (exceeded max delay comp even on long ADC)
region length = 137591
snippet length = 2820
starting location of snippet audio source:
from start of region = 3701
before beginning of snippet = 137591 - 3701 - 2820 = 131070

Recordings 2-5 were in a session with lots of tracks and plugs. To do Recording 1, I created a brand new session, added 2 audio tracks, put a plug (Waves Linear Phase EQ since it has a long delay) on track 2, recorded on track 1 with ADC on. No other plugs, no weird routing, same result. Recordings 3-5 also had a Waves Linear Phase EQ in there to create a large delay and thus larger snippet (makes it easier to find the snippet in the recording if it is larger). To be sure it wasn't a Waves problem or dependent on really long delay compensation, in Recording 2 there were no Waves plug-ins and a relatively low total dly+cmp, but same results.

Digi, I can provide the audio files if you like. I examined and listened to the audio files in another program to see if it was playback-related but the snippets do appear in the WAV file even when played in other programs.

Hopefully this will point you in the right direction. PLEASE FIX THIS!

Let me know if I can provide any further info.

Respectfully,
Your friendly neighborhood customer/QA Engineer

Re: DIGI-More observations on click/pop problems
 

Hi,

I will forward this information along immediately. Thanks for the obvious extra effort.

Digidesign Tech Support

Re: DIGI-More observations on click/pop problems
 

While I didn't do any brilliantly exhaustive tests like AJS did (thank you!), I can confirm a few aspects of his observations and add a few.


1) The audible disturbance is indeed material previously handled and cached by the session.

2) Happens to me when selecting a time range (either on a track or a ruler) and then "printing" to a track.

3) I can confirm that if the selected time range includes a long-enough silence at the end, the problem does not occur (I realize it may still be occurring, with the system now "dumping" cached silence).

4) Only happened to me using PT 6.4 cs8 and onward, (OSX 10.3.4 and 10.3.5). Never before.

5) I use ADC at all times.

6) Happens regardless of the session's file format (mine are mostly SDII or AIFF).

7) It happens also without enforcing cross-platform compatibility (I never do).

8) My system does feature a Sync I/O. The problem happens whether the clock master is a 192 or the sync I/O.

9) My system does feature digital connections (see below).


G4 933 Quicksilver
1.2 Gig RAM
OSX 10.3.5 (was 10.3.4 until release of PT6.7)
System drive: 120 Gig Western Digital 7200 RPM, 8MB cache (Formatted with Apple Disk Utility - 2 partitions, 60 Gig each)
Digi/Atto SCSI 128 Dual Channel Host Adapter (firmware 1.66)
Atto Configuration Utility 2.6.1
StoreCase ULVD3 SCSI drive bay (4-slot)
Seagate Cheetahs 15K (formatted with Apple Disk Utility
PT 6.7 TDM
Sync I/O (firmware 1.1.0)
3x 192 I/O (standard; no expander cards)
Digital connections (AES/EBU):
TC 6000 (x4)
Orville-R (x2)
TC Finalizer 96K (x1)
Tascam DS-M7.1 (x4) (w/ both A/D connections)

I hope this helps.

Re: DIGI-More observations on click/pop problems
 

Regarding the problem of the snippet of audio that occurs at the end of a region when recorded with ADC is on... Here is some more info:

First of all, the problem only happens (for me at least) when no range is selected (the start and end point are the same) and I manually stop the recording. If I select a range, then the snippet does NOT occur even if I manually stop in the middle of the range during recording... ie, it works properly (Strange that someone posted the exact opposite of this a few months ago... don't know why).

Second, as I mentioned previously, the snippet that is stuck at the end is the audio that is located about 3 seconds prior to the end of the region (ie, 131072 samples prior to the snippet start). HOWEVER, if the recording is less than 3 seconds in length, the snippet comes from audio recorded in a PREVIOUS take. If it is the first recording after opening a session, then the snippet is blank.

Try this:
1. start PT, create new 44.1/24 session with 2 audio tracks, put a plug-in(s) with a reasonably large delay on the 2nd track, turn ADC on, rec-enable 1st track, select a start point (but not range), connect an instrument, let a note ring, hit record then hit stop within 2 seconds, all the while letting that note ring. Zoom in on the end of the region. You should see a range equal in size to the dly+cmp where there is silence, even though the note was ringing.

2. Now hold a note and record >3 seconds of audio, holding the note the entire time (again, do NOT select a range just a start point), then undo, silence your instrument, then record 2 seconds (or any length less than 3 seconds) of silence. Zoom in on the end of the supposedly silent region and you will see a snippet from the >3 second recording.

HERE'S MY THEORY:

PT is using an approx 3 second circular buffer to store the audio before writing to disk. Being "circular", when PT gets to the end of the buffer it starts writing to the beginning again. It reuses the buffer without clearing it between takes, leaving old audio in the buffer.

When ADC is on, PT is assuming the musician, who is playing to the delayed signal, is playing notes that are late by the max amount of delay (call it d samples), so when it reads from the buffer to write to disk, PT reads from sample location that is d samples later than the time it is writing to disk.

BUT... when a user who has sets a single start/end point hits STOP during a recording take, PT stops recording to the buffer immediately rather than recording another d samples. Since these d samples are written to disk at the end of a region, it is writing samples either from the previous recording take, or if the current take is longer than the 3 second buffer length, from 3 seconds prior to the end of the current recording!!! So if PT continued recording to the buffer for d samples after pressing STOP (as it should since the musician should be hearing everything d samples late), it wouldn't have the snippet from the previous recording left in the buffer.

ok, this is just my theory and I don't know anything about how the PT code works, but desperate times call for desperate measures!

Re: DIGI-More observations on click/pop problems
 

Digi,

Since there is no mention of this problem being resolved in cs2, I was wondering if you have been able to reproduce the end-of-region audio glitch problem following the instructions in my last post on this thread.

Re: DIGI-More observations on click/pop problems
 

Yes, what is the status of this bug?

Has it been flushed out of PT 6.7?

If not when will it be?