Click detection

This forum is for discussing Wave Corrector, audio projects, tips & tricks etc.

Moderator: Derek

Click detection

Postby Michel » Tue Jan 11, 2005 3:10 pm

Hi,

Happy new year to all. And congratulations with this new forum!

I am one more happy user of WaveCorrector. I use it mainly to clean up unique recordings on vinyl records, before burning them on CD with Feurio (another great program. You should try it, if you don't already!).

What pleases me most is the controll you have on the corrections made by program. Other programs do not show the individual corrections, and the only control you have is setting parameters and/or the undo button afterwards.

However, the parameters of the click detection and corrections algorithms are quite few in WaveCorrector. At least those that I found! (are there any more?): Detection level 1..5, and differential/per channel.
Most of the time, level 2 or 3 is best for me. But sometimes level 2 leaves too many clicks uncorrected, while level 3 gives thousands of corrections.

Does anyone recognize this?

Most of the corrections have 'value' 1 or 2. They may be very light crackle or false detections. But if they are so many, they are degrading the sound quality, rather than improving. The light crackle is replaced by a kind of rumble in the lower frequencies. I find this less pleasant, and there is very little you can do to it afterwards.

A solution could be to remove all the corrections with value 1, leaving only the bigger ones (the real clicks), but I see no way to do it, other than removing 5000 corrections manually.

Another problem I have with the click detection, is that some instruments tend to cause a lot of false detections. These are some percussion instruments, and also plucked double bass. I wonder if the click detection algorithm could be 'tuned' in those cases, in order to distinguish the typicel transients of these instruments from scratches on the record.
One can do it easily by looking at the waveform...

So far my experiences with WaveCorrector. The other features (track separation, chanel balance, tone control) I hardly use.

I am looking forward to read your experiences!


Regards,
Michel van der Mark
Royal Conservatoire
Den Haag, NL
Art of Sound dpt.
Michel
 
Posts: 1
Joined: Tue Jan 11, 2005 2:34 pm
Location: Gouda, Netherlands

Re: Click detection

Postby Derek » Wed Jan 12, 2005 11:22 am

Michel wrote:... the parameters of the click detection and corrections algorithms are quite few in WaveCorrector. At least those that I found! (are there any more?): Detection level 1..5, and differential/per channel.
Most of the time, level 2 or 3 is best for me. But sometimes level 2 leaves too many clicks uncorrected, while level 3 gives thousands of corrections.

A number of users have asked for intermediate settings between the current 1 to 5. However, It is difficult to do because click detection/discrimination is far from perfect. It would be nice to think that if you had an intermediate setting between say 2 and 3, it would detect all the clicks missed by setting 2 but reject all the false positives detected by setting 3. Unfortunately, I haven't really been able to achieve this. Nevertheless, be assured I will keep working on it.

Another problem I have with the click detection, is that some instruments tend to cause a lot of false detections. These are some percussion instruments, and also plucked double bass. I wonder if the click detection algorithm could be 'tuned' in those cases, in order to distinguish the typicel transients of these instruments from scratches on the record.
One can do it easily by looking at the waveform...


Yes, it it not easy to create a computer algorithm to do what the brain does effortlessly (discriminate between a musical waveform and a noise spike). This is something I've worked hard on with some success. However, there are still circumstances where the algorithm fails. Although it is possible to tune the algorithm to accept/reject particular waveforms, this usually results in the algorithm failing when presented with a different type of music. The current algorithm represents my best compromise between false negatives/positives.
Derek Higgins
Wave Corrector Developer
http://www.wavecor.co.uk
Derek
Site Admin
 
Posts: 476
Joined: Sat Jan 01, 2005 7:57 pm
Location: Coleshill, United Kingdom


Return to Chat

Who is online

Users browsing this forum: No registered users and 1 guest

cron