searching for a repeated series of samples

GoldWave general discussions and community help
Post Reply
Gord
Posts: 391
Joined: Thu May 21, 2009 4:26 pm
Location: Canada

searching for a repeated series of samples

Post by Gord »

Hi, all. I'm a brand new user - registered today - and I have been very impressed with what GoldWave can do. I have also learned a lot through trial and error, and by reading the online help and searching the forum, but I know that I have just scratched the surface. Here's my issue:

When I record from tape or vinyl I find that once or twice in each song the sound will stutter. I found an old forum topic here

http://goldwave.ca/forums/viewtopic.php?t=1011

and tried the suggestion of setting Temporary Storage to RAM (I have 1.5GB), but that did not alleviate the problem. I also updated my sound card driver, BIOS, etc..

On reading some other posts it seems like my Realtek sound card might be the culprit. So, I'm looking for a workaround.

It seems that each "stutter" is an 80ms series of samples that is repeated twice. I found this by looking at the waveform and noticing the same peaks and valleys. When I selected between two similar points in each section the length of the selection was always 0.080 seconds.

So I was wondering if there was a way I could get GoldWave to search for this repeated series of samples and list them for me instead of my listening to the tracks in real time with my finger on [Ctrl-Q] to mark where the hiccups occur.

Any advice would be greatly appreciated!
DougDbug
Posts: 2172
Joined: Wed Feb 16, 2005 3:33 pm
Location: Silicon Valley

Post by DougDbug »

I don't think there's a way to search for repeated samples.

Yeah... You've either got a driver problem, or maybe something's running in the background that's hogging the CPU or databus, and causing the input buffer to "run dry".* Sometimes an antivirus program can slow things down. I assume that you are not intentionally running any other applications while recording. There is a good chance that you're missing data where the audio is repeated. (This may, or may not, be audible after removing the glitch.)

This really shouldn't happen, and if you can't solve the problem you might need to try a different soundcard.

Normally, when everything's working correctly, the recorded sound quality is not affected by the software (or drivers) at all. The analog part of your soundcard and its A/D converter can have an effect (you can get noise or clipping), but after that it's just a digital data stream. As long as that data stream doesn't get interrupted, there should be no alteration of the data (until you intentinally alter the file with GoldWave).
instead of my listening to the tracks in real time with my finger on [Ctrl-Q] to mark where the hiccups occur.
:lol: Audio editing tends to be time very consuming. :lol: Depending on what you are doing, you usually end-up listening to the full program several times. Right now, I'm fixing-up (de-clicking) a vinyl transfer. The record is in poor shape. I've already spent a few evenings on it and I'm not done with the 1st side yet! This isn't unusual. I usually spend a day or a weekend on a project like this, listening to the recording (small sections at a time) over-and-over!

Slightly off topic - If you want some information on removing "snap", "crackle", and "pop" from vinyl-to-digital transfers, this page has tons of information. And if you're interested, I can post an outline of my "standard procedure" for vinyl transfers.


* The record/play buffers are usually "circular", and you can get repeated sounds when something goes wrong.
DewDude420
Posts: 1171
Joined: Fri Mar 11, 2005 11:15 pm
Location: Washington DC Metro Area
Contact:

Post by DewDude420 »

currrently there is no way to do this automatically, you're going to have to do it manually.

it sounds like a daunting task..but..doug is right..audio editing is VERY time-consuming...i've spent HOURS on just seconds of audio.
Gord
Posts: 391
Joined: Thu May 21, 2009 4:26 pm
Location: Canada

Post by Gord »

Thanks for the replies. I isolated a couple of the "stutters", and took a closer look. Although they all appeared to be exactly 3528 samples long and they looked the same in GoldWave's Sound window (even when I zoomed in *really* close) when I dumped the numeric values of the samples and compared them in Excel they are not the same. They are similar (mostly), and they move in the same directions (mostly), but they are variable enough that it would probably be tricky to write a little program to match them up.

I was hoping that the glitches were introduced after the analog-digital conversion and the individual sample values would be the same. If that was true then the code to find the duplicated sections would be easy.

BTW, I also tried a recording after using msconfig to disable all of the startup items (anti-virus, etc.) and the stuttering still happened. :(

My next step is to try recording on a different computer.

p.s. Doug: Thanks for the link! I also have a question about the order of applying effects so I'll start a new thread for that.
GoldWave Inc.
Site Admin
Posts: 4372
Joined: Wed Mar 10, 2004 6:43 pm
Location: St. John's, NL
Contact:

Post by GoldWave Inc. »

I cringe whenever I see the word "Realtek". My advice is to get better sound hardware. That will save you a lot of trouble and give you better quality in the long run.

Chris
fcampbel
Posts: 7
Joined: Tue Jun 02, 2009 8:38 pm

Post by fcampbel »

I've been using GoldWave for several years and really like it. I've never really understood all the technical details and math but I frequently use the clipboard to reduce background noise. It seems to me that the it works like this: you select a portion of the wave form and copy it to the clipboard. When you tell it to, the program searches through the complete file, looking for matches with the data in the clipboard. When it finds a match, it removes it.

Recently I decided that I wanted to search for the clipboard contents to place a cue point there instead of deleting it. But GoldWave can only do this for silences. Why can't it have an option in the automark function to look for the clipboard contents rather than silences?
Fred
DewDude420
Posts: 1171
Joined: Fri Mar 11, 2005 11:15 pm
Location: Washington DC Metro Area
Contact:

Post by DewDude420 »

fcampbel: you're half right on the noise reduction.

noise reduction is an FFT process and works on the frequency domain and not time. when you copy audio to the clipboard, the NR filter analyzes the audio in the clipboard basically converting it to frequency domain and then plots there in the frequency spectrum the noise occurs...when it's analyzing, it's basically marking.

after it's read the clipboard audio and has a spectral-readout of it, it then proceeds to process the actual audio in the frequency domain, reducing the spectrum by the amount specificed by the noiseprint in the areas where noise is...so for example, if you had a lot of rumble below 200hz but there was no noise above it, then the plugin would only modify the frequencies below 200 hz...likewise, if you have constant white noise that's got even distribution across the spectrum (commonly called a noise-floor), then it would reduce everything in the spectrum by whatever amount is in the noise.

sample comparison would have to be done in the time-domain as frequency-domain matching, while probably very accurate, would be time consuming and processor intensive...however, it would allow a bit of "error" room in case the same sound later in the file doesn't have the exact matching waveform...

in fact, come to think about it, this kind of deal would have to be done in the frequency domain to have any kind of accurateness to it.
Post Reply