Mystery nuisance click sound on exported wavs

[velvetcave]

Hello,
Apologies for a long post.
I have a mystery click sound on audio files that I consolidate in a pro tools session and export as 24bit 44.1kHz wav files, I`m hoping someone here might be able to explain it.
I run pro tools 5 on a G4 with os9 and have done without a major problem for many years. I record multitracked string parts as a remote session player and send them on to my clients as raw consolidated and exported wavs for them to include in their session files. I have had a problem with nuisance clicks appearing in waveforms that have been consolidated in the past, but it was very rare, always visible in the waveform at my end and solvable by simply consolidating the audio again. This time, it has become much more frequent and inaudible and invisible in my session, even when the client sends me the files back and I load them into a new session. Also when I listen to those files on a bog standard PC laptop I don`t hear the noise. I have thought about clipping at the client`s end but I`m getting this problem with multiple clients, some of whom have been using my stuff for years and are commercial media music producers so I think it`s unlikely.
Here`s my method if it`s any help:
I use a session file that I have stored as a template containing all the tracks, routing and subgrouping I usually need.
I "save as" for the project I`m working on setting the sample rate and file allocation appropriately.
I record the parts.
I edit the parts, this involves topping and tailing each discrete bit of audio to remove unwanted noise, space bar and mouse clicks etc, I usually put a small fade at the beginning and end of each bit of audio within a track and occasionally splice audio within a track and place crossfades between parts, when tidying up staccato for example.
I then consolidate the tracks to produce continuous audio files for the client to drop into their session.
I export the files as wavs, always 24bit, sometimes 44.1kHz, sometimes 48kHz.
I transfer the audio to a LaCie drive using firewire then transfer the files onto a windows PC for internet connection.
I then zip and upload the files, usually to Wetransfer, then send the link to the client (the problem also happens if I don`t zip).
I`ve been doing it this way, every day for years and the problem outlined above was incredibly rare. Not so now!
One interesting aspect is that the problem seems to be confined to certain tracks. The template file is arranged like an orchestral string section, so the first eight tracks are for cello, the next ten tracks are labelled viola, then 12 each for violin2, violin1 etc. I only ever get this problem with the viola parts, but not always the exact same tracks in the session, just within that group of tracks. This makes me wonder whether there may be some corruption creeping in to the template file, so I built a new one from scratch. I`m just about to test that out but if anyone here has experienced this before or might be able to suggest a reason why this is happening I`d be very grateful.
My system:
PowerPC G4 1.25 GHz
2 Gb ddr sram
os9.2.2
pro tools v 5.1.1
882 [20] io
Core and farm card
The only material change to this recently was the addition of a couple of new drives about nine months ago that I use to store the session files and associated audio. The template exists on an old drive I`ve been using for about 5 years and the operating system is on a separate drive altogether.
Thanks for reading!

[velvetcave]

Update, just checked the files from the new template and they appear to be clean, however I have yet to get feedback from the client.

[Philgood]

I had once noise problems on my AKAI S5000 and it was down to the RAM so check that also on your machine. Maybe bad ram is corrupting audio ?

[velvetcave]

Hello Philgood,
Thanks for the suggestion. I replaced my ram about 18 months ago, but the cards weren`t new so could be. Just downloaded the diagnostics program from here so we`ll see.

[Philgood]

If you have some spare RAM there just try it out i guess. Despite of whatever is the outcome of the tests.

[Philgood]

I also have a complete Protools system running here on OS9 so that could come in handy ?
Maybe i could crosscheck some files for you.

[Andre Solomko]

Hello, i would also think about PSU. Thing is that capacitors are slowly dying and it might affect your sound in a amisterious way.. I recently checked  m audio USB card which i  previously tried as additional SPDIF unit. Had a horrible noises so i abandoned it . Replaced PSU with an new ATX and tried card again. Works fine. So yes it might be RAM or Hdd. And I have the same machine as you i guess, only PT LE .MDD are notorious for PSU failures. Its age but they were not so  good at beginning. Last thing: why it affects some group of tracks only...After all its a DSP system and what I learned that Analogue word failures can be seen in a digital world by some really exotic way.

[velvetcave]

Thanks for the offer Philgood, I may just do that if this fix doesn`t solve the problem. I will be swapping out the RAM cards when I get to it next week. Always the problem with random and intermittent faults though, you can`t be sure it`s worked until months later!
Interesting suggestion Andre and thanks for that, I have a couple of spare PSUs so I`ll try that too when I have it in bits. It seems to be on consolidation and export, I check all audio thoroughly before this stage and obviously there`s never a problem. RAM seems more likely? Or a card failiure? Or corruption on the HDD that the template is stored on? I`m currently looking for replacement core and farm cards but the right ones don`t seem easy to come by.
Also I`m conscious there are a lot of processes for the files to go through before they reach the end user. Firewire, USB, zip, upload and download could all be potential sources for error? I`m starting to wish I knew a lot more about the processes.
Meanwhile I`ll pursue your suggestions and do some diagnostics on the disk and RAM.

[GaryN]

You are finding yourself in the middle of a logical troubleshooting issue. Am anomaly is appearing so you're looking for a cause.
There are things you MUST determine with certainty as you do this or you will be led down wrong paths or just around in circles.
By this I mean compiling what you know to be facts and separating those from what are yet to be proven.

Example: "I have had a problem with nuisance clicks appearing in waveforms that have been consolidated in the past, but it was very rare, always visible in the waveform at my end and solvable by simply consolidating the audio again. This time, it has become much more frequent and inaudible and invisible in my session, even when the client sends me the files back and I load them into a new session."

This would clearly say that something wrong was happening in the consolidation process, since repeating it would clear it up. But NOW, they're "much more frequent and inaudible and invisible in my session".

You need, at this point, to confirm that this is actually the same thing happening and not an entirely different issue. because you've had a "click" problem in the past and now multiple clients tell you they hear clicks in your product, you're assuming it's the same problem. However, you can't see it or hear it. This is where everything goes wonky. Logically, they must still be there.

So, start by examining some of the problem files in an audio editor…NOT in PT. Confirm you can see the glitch in the waveform. You want to certify that you're actually sending glitched consolidated files. The real mystery here is that they "disappear" when they return and you dump them back into a session. If necessary, get one of your best clients you're on great terms with to mark some of the clicks they hear so you have some exact or near-exact places to look. It defies logic that you can send a "clean" file, have it be "dirty" on the client's end then have it magically repair itself when they send it back. See if you can correlate those clicks with punch-ins or other pasteovers etc.

The other mystery…that the clicks only appear in viola tracks would seem to be screaming that there's a template problem, though what it could be I have NO idea. This however, is a critical point in all of this. ALL of the other speculation about file transfer-over-internet caused corruption is moot if it only happens to certain specific tracks.

You definitely need to create a new template for all of this…from scratch… Using any part of the old template can continue the uncertainty of it possibly being the cause if the problem continues. All of the speculation about PSU-caused glitches is relevant ONLY if you can see the glitches in the waveforms because they were disrupted by a voltage spike or such during record leaving a permanent, visible glitch.

There ARE other possible causes for digital audio "clicks & pops" - especially when a LOT of tracks are playing, These range from insufficient buffer size (because we all hate latency), to an HDD NOT being in DMA transfer when it should be, to just running too damn many plugins. THESE however, will cause clicks at random in playback and NOT get written to the files. So, anything like this goes out the window if you can SEE the glitch written permanently. Think hard about anything you might be doing recording or processing the viola tracks that's different from the other instruments. Maybe punches are NOT happening at zero-crossings where they should be or something like that.

There's an answer here somewhere.

[velvetcave]

GaryN, thanks for your detailed and very logical reply. I think you`re spot on and the conclusion you came to about the template is the same one I came to, I suppose when stuff like this happens it`s too easy to get sucked into other logical worm holes at the same time, whereas I know that it`s always best to try one fix at a time. Also I`m aware that the system is quite old and I have a lot of processes to go through whilst transferring. Given that my first recording device was an old ferrograph reel to reel it is still a source of wonder to me that what I do now works at all. But we trust... until something goes wrong! Again you`re quite right, if the transfer process were to blame it would probably be much more random.
I have built the new template from scratch and it now sits on a nearly new HDD.
"compiling what you know to be facts and separating those from what are yet to be proven." This has thrown up the root of this anomaly, the client who did send these files back has a system I`m not familiar with and is not necessarily to be trusted regards doing things like adjusting buffer size. I don`t doubt that he did receive some corrupt files but I also suspect that he`s trying to squeeze too many of my tracks into a "small" set up. So when I fix his files he still has a problem. When I send entire string sections as individual exports there are a fair number of audio files, in this case 24. This could well be more than his system settings would be comfortable with. He`s heard the glitch in the violas before so naturally he starts removing them from the session and bingo, the problem disappears. A classic double glitch.
"Maybe punches are NOT happening at zero-crossings where they should be or something like that." This sounds interesting, could you explain please?
So I had a first run with the new template and all seems to be ok, I`ll be doing another 4 or 5 in the coming week so hopefully they`ll be ok too. Thanks again.

[GaryN]

I have thought about clipping at the client`s end but I`m getting this problem with multiple clients, some of whom have been using my stuff for years and are commercial media music producers so I think it`s unlikely.
*********
the client who did send these files back has a system I`m not familiar with and is not necessarily to be trusted regards doing things like adjusting buffer size. I don`t doubt that he did receive some corrupt files but I also suspect that he`s trying to squeeze too many of my tracks into a "small" set up. So when I fix his files he still has a problem.

This is exactly what I mean. These statements are at odds with each other. It does sound like the new guy is just running out of CPU. Find a tactful way to tell him to try mixing them down into fewer groups.

So I had a first run with the new template and all seems to be ok, I`ll be doing another 4 or 5 in the coming week so hopefully they`ll be ok too. .

It sounds like you're one very busy one-man string section! Let's hope it's a cure. Holla back.

[velvetcave]

I certainly will, I`m coming to the conclusion that this is a double glitch. Multiple clients experienced the first issue but it`s only one client with whom the problem persists, Hallelujiah!!
I did as you suggested and looked at the returned waveforms in an audio editor I use on another machine, no click, either visible or audible. I have tactfully advised the client to increase his buffer size so we shall see.
The new template is very much from scratch, the only thing it shares with the old is an altiverb plug in I use for demo bounces. This obviously never figures in the exports. Indeed it struck me whilst I was creating it that it was the first time I`d launched pro tools independently from the templates I use for a good number of years.
Remote string session work is pretty much all I do now, hence the panic. I used to do a lot of live work and touring but I don't have the inclination these days.
I`ve come to love my old Pro tools system, it`s clear and concise and easy to use, and normally rock solid reliable. There`s always that nagging doubt though that some day it`ll fail and I`ll have to upgrade. I shudder at the thought.

[GaryN]

Hallelujiah is right! Glad to hear it!

Note: "the only thing it shares with the old is an altiverb plug in I use for demo bounces"
I'm not familiar with exactly what PT does during consolidation BUT could leaving Altiverb on during that process (maybe by accident) possibly cause a problem? i.e. you're smashing parts together into one stream and there are unnoticed reverb or ambience tails that get chopped off?…just a thought

"There`s always that nagging doubt though that some day it`ll fail and I`ll have to upgrade. I shudder at the thought."

The time to "future-proof" hardware is: 1) While you CAN still find it…and  2) While you're making $$ and can afford it.
I don't have the proprietary Digidesign hardware issue since i don't use PT but I DO have spare everythings for all else. Interface, computer parts (by buying whole G4 MDDs) and everything else that has no substitute that could stall everything on failure. It's no different than backing up software…just more expensive…

Ciao…

[velvetcave]

Ok, after using the new template for a week on 5 different projects all is well. Consolidation of tracks produced no visible clicks and I`ve had nothing back from my clients to suggest there`s a problem. It`s interesting though that this was a glitch in the template that seemed to accumulate over time. Having used the same template written to the same disc for at least 5 years maybe points to some degradation there. Who knows, but I`ll be rebuilding it periodically from now.
I`ve also convinced my client with the double glitch to have his piece mixed and mastered professionally by someone I know well, they also report no problem with the files.
Many thanks to all who offered advice, particularly GaryN who reminded me of the logical approach. I`ve got most of the components together for a complete replacement system, the only things missing are the mix core and farm cards which seem to be scarce on ebay just now. If anyone knows where I might get hold of these I`d be very grateful.
Thanks again:)

[mrhappy]

Hey velvetcave... glad you got things sorted out! Gary is always the voice of reason when things start going haywire!! Haha! For some  reason i never saw this post or would have joined in the fun! This is moot at this point but I'll throw a few things out there anyway.

Usually when some little oddball problem like this arises in PT they recommend that you trash your pref files. It's kind of like flushing the PT toilet and will rebuild upon startup.

Like Gary mentioned... It can be worthwhile to make sure that your punches/ edits take place at the zero-crossing points to prevent clicks etc. Fades and crossfades certainly help here however they just 'cover up' the clicks so you don't hear them.  On a heavily edited track containing tons of crossfades/fades I never COMPLETELY trust the consolidation process and usually prefer to bounce to a fresh track where I can hear whats happening in realtime. Is it possible that your viola tracks contain the most editing?? Maybe that's why your problems showed up within the viola section?? Just a guess!

They used to advise that you open the PT program first THEN open your session/ templet instead of launching the program by clicking on session file... don't know if that's B.S. or not but I've been following that advise since.

Anyway... Seems all good now. The kids around here are quite helpful! Would love to hear more about your whole 'String Orchestra' process!!