Maybe serious lack of sleep here, but I'm looking all over and I don't see anything like it in Sound Forge. Don't know if it would help as the clipping seems to have happened in the DAT even though the audio never peaked above -8db and it doesn't go above that in Sound Forge or Vegas either.
Any help is greatly, greatly appreciated.
Those FX's are additional options that don't automatically come with every copy of Soundforge. I happened to purchase NR 2.0 which in that case included Sound Forge.
I'm sorry, I ment I have Sound Forge 7 and NR 2, but I can't find the clipped peak restoration tool. Is that an add on to noise redux? Would it even help me?
Clipped peak does come with the NR package... so you should have it as well. All the FX's are called Sonic Foundry XXXX e.g. Sonic Foundry Noise Reduction, Sonif Foundry Clipped Peak.
It should be under DX Favorites/Sonic Foundry (along with quite a few others that came with NR 2)
Theoretically it should help (I think)... and I am in an identical situation myself with a wireless MIC someone used in a recording where the mic was not attenuated with the transmitter and the signal was peaking. The audio, like yours never reaches above -3db in the file and yet the audio is distorted. Zooming in using SF you can see the lovely square peaks (strangely always on the lower side of the "wave").
I tried using the clipped peak tool and it seems to make little diferrence (which could mean I am using it wrong... or my ears just aren't that sensitive..and/or the audio is just too far gone for the FX to recover it). There is a reference in the help which says the FX is only for "occasionally" clipped audio and that if the audio is excessively clipped that it will not help.
If you figure it out and your audio is improved.. let me know as I would love to be able to "fix" this audio as well.
I've used the NR 2 pack in Vegas (only have that el-cheap SF that came with Screenblast Acid & it doesn't support plugins). The clipped peak tool will work on abunch of waveforms at once, but they can't be too far gone. I've used it to fix clipps in recordings from the 50's on reel to reel. Works good.
If eighter of you haven't, read the PFD help file. It REALLY helps. :)
Thanks Liam and TheHappyFriar. I misunderstood the help commands in SF and was opening the noise reduction plugin looking for a setting in it rather than a seperate plugin. Not quite sure why mention the cd when that's not the place to look. But then again, I haven't slept since thursday, so it might be me.
I've also checked my waves in SF and there is definitely clipping, even at -12db. I have no explanation for that. The level on the DAT I recorded with was set to about 1 (out of 10).
So far the clipped peak thing isn't helping, but I need to more research. Any tips are certainly appreciated as are explanations why my audio clipped at such a safe level.
I'd need to know HOW the recording was made to tell you how you got clipping. But in general even though the recording devices isn't showing clipping it can happen before it gets to the recorder.
Just as an example, wireless mics have a transmitter, now if the mic level is way too high then the front end of the transmitter is going to clip the signal right up front. Conversely if the level going in is way too low you'll get more noice than you should.
Same thing can happen at the reciever end, if you're coming out of there too hot you'll overload the front end of the next stage in the chain.
I think the Sennheisers have a level meter on the transmitter, checking that you're getting the right sort of levels there can really help.
Also you need to know what meters are reading, mostly these days they seem to read peak but you never know!
Thats why it always pays to monitor with closed back headphones when you're recording, better still do a test recording and listen to it carefully. Track I find when listening is to close your eyes and press the cans hard against your head to get the brain really focused on the sound from the cans. If the sound in the cans is bad but the ambient good it seems our brain just blocks out what its hearing from the cans and this seems to be more so when we are getting visual cues.
I think the trick with clippered peak restoration is telling it at what level the peaks that are clipped are at. . I've only played with it briefly and the clipped detection tool as well, you can use that to see if you've got things setup right and then pur the values you used there into the restoration FX.
Sounds like something was clipping the signal before it got to your DAT machine. What was your signal chain? Mic into a mixer first? Right into the DAT? The problem with clipping in digital...once it's clipped, it's clipped. As apposed to analog where a little clipping was good, gave a natural tape compression. Not all pieces of gear have level meters, so you really gotta be careful when setting levels in the signal chain.
It was mic directly into the DAT. The mic was a Sennheiser ME66 and the cable was 25'. I don't think the cable ran over any electric wires, if that helps.
Any thoughts on what could be the problem? Is there any way to use the pencil tool to fix things if it comes to it?
Thanks again for the help.
You can use SF to find the clipped peaks, I seem to recall it marks them as regions and then you could use the pencil tool to draw a rounded top on them to at least get rid of the harmonics. That would be a HUGE effort though, I think I'd be spending a bit of time getting clipped peak restoration to do the job.
Hmm ME66, think that's the mic you can either put a battery in OR run off phantom power. If it's the one I'm thinking of then having the battery in and phantom power on causes no end of problems.
If it's not the one, you could have succeeded in simply overloading the mic.
May be really stupid question, are you 100% certain it's clipping?
I guess if you're that far into it your pretty damn certain, just that I've fretted over what sounded like something clipping but it wasn't, just the way the guy's voice sounded.
Come to think of it you mentioned only the negative peaks were clipped, hm, maybe something wierd was happening in the mic, speech oftenly has more negative than positive peaks but for it to be enough to cuase clipping in one direction is odd. You can use SF to see if you do have a DC offset, I though those problems only afflicted very old A/D converters.
Thanks for interest and help. Sorry I didn't respond earlier. Didn't sleep thursday night getting these things ready for the festival.
I double checked and the clipping is at the top and bottom of the waves. The ME66 didn't have the battery power running but was using the phantom power from the DAT. The DAT's limiter was not on. So again, I really don't get it. Maybe I was wrong and The mic cable intersected the power cord of the DAT as it was not running on battery. Don't know if that makes a difference.
I'd love to post a section of the clip on my web site, but I only know how to use the basic builder that comes with the hosting, and I don't think it allows audio to be added. I could email a 5 sec bit of it or create an email address and pu it there and post the password.
Would anyone be willing to put up with that?
Thanks again for all the help!
I wouldn't say it's pristine but to my jaded ears its a lot better.
Issue is as I discovered when I had a play with the clipped peak restoration tool is it only works on peaks OVER 0dBFS.
In this case the clipped peaks were a LONG way below that.
So trick was, first nomalize, the add just enough extra gain (using Volume in SF) to get the peaks JUST high enough for the FX to see them. The peaks should have nice flat tops.
Use CPR (Ha, never thought of that before, nice one SoFo) with Crossfade Edges and 6dB attentuation to give SF enough headroom to rebuild the peaks into.
The clip you (musman) sent me didn't look like it had any clipping in it at all (even when i normalized it). None of the waveforms looked flat (on eigther the top or bottom).
What it SOUNDS like could of happened (because when the guy says "reality" it shoulds harsh) is that he spoke to close to the mic. I hate it when that happens (that and the "POP" sound. :) )
I'd agree the guy has a harsh voice but on top of that there's definately some clipped peaks. After I processed it in SF it still sounds harsh but the clipped sound is out of it. A bit of compression to flatten the level out would also help.
Maybe he'd sent you a mp3? The mp3 compression may have done something to it.
The multiband dynamics has a preset that's fairly good at getting rid of the plosives, still better to get the mic further back.
Thank you both for looking at it. Wow, the new version is definitely better. It's still not perfect, as you mentioned farss, but a lot better. I'll try to tweek it from here. Thanks so much for the settings, farss.
I'm still at a loss to how this happened. The mic was probably 3-4 feet from the actors and the levels fine. I thought I heard a bit of clipping when we were recording, and I should have done more about it. But getting a mic further from the talent is not something that's ever been an issue and I didn't know you could have a problem like this. I'm guessing this was the problem as I've no other solution. This is why my first movie was silent.
Anyway, thanks again for all the help. I do appreciate it very much.
Glad to be of service.
I think you'll find that's a very sensitive mic so at the distance you are working at there might be a fair chance of something getting overloaded. It's possible there was too much level for the front end of the mic preamp so even winding the gain down wouldn't have helped. Probably not a bad idea to have one of those XLR 10dB pads in the kit although you might have to be careful how it sits with the phantom power.
Also it seems the ME66 has a low cut switch on it, that may have helped a bit too, the clipping seems to be mostly at the low end and in that close with that kind of mic you're probably getting more bass due to the proximity effect than you'd need anyway.
It's pretty good. It gives mic recomendations, how to use them, the history/science of them. It might tell ya what went wrong so next time you know what to avoid (i know I do! :) )
Kudos to the first person to find it: i forget and don't feel like looking (i just got up!).