0% Interest for 24 Months! Learn more »
(800) 222-4700
  • Español: (800) 222-4701
June 2017 Giveaway

Sweetwater Forums [Archived]

After 15 years of great discussions, the Sweetwater Forums are now closed and preserved as a "read-only" resource. For discussions about current gear, check us out on Facebook, YouTube, inSync, and our Knowledge Base.

How to eliminate "tape hiss" in an audio CD ?


I just bought an audio CD from Amazon.com . While playing it, I hear a lot of what sounds like "tape hiss" in quiet spots. It's the same sound that used to be eliminated with the early Dolby filter for cassette tapes.
How do I eliminate the hiss? Is there some "free" way to copy the CD while removing the hiss from the copy?
February 7, 2006 @11:33pm

The answer is that it is often possible to make significant improvements but I don't know how far you can get for free. The process is called noise reduction or hiss reduction. While hiss reduction software is especially oriented to that particular kind of noise (i.e. 'hiss'), it is in fact often easier to get better results without the specialized approach, by just using noise reduction.
Noise reduction in general is a facility that comes with audio editors. Audition, for example, has good noise reduction facilities. In many cases, especially with tape hiss, I can get better sound with the Sonic Foundry Noise Reduction DX plugin. This works from within an audio editor, such as Sound Forge or Audition; you need the editor in order to use the plugin.
Nero has a hiss filter that can be applied to tracks befor it writes them to CD-R. I have no idea how good it is, but if you already have Nero, the filter is "free." Some of the freeware editors, such as Audacity, have some form of noise reduction. You might find other freeware or shareware especially for this task if you do a search.
I suspect in this regard you get what you pay for. It is not particularly difficult to remove a lot of hiss; what is hard is not destroying the vitality of the music at the same time.
February 8, 2006 @09:26pm

I tried what you describe and you are quite right. You can remove a lot of the hiss but you get a little something else back in return but it's a major improvement!
February 9, 2006 @12:20am

I am sure there are better programs out there, depends on how much time
you have.
I had a problem with a bass player that sent files using zoom pedal effects
that had a boatload of noise with them. I messed with parametrics, whatever.
I ran the files through Emagix program designed for copy-ing tapes and records. I used the 'dehisser' within reason and got pretty good results, granted the his is still there,
but I have less parametrics to deal with.
Emagix audio clean up lab sells for like 30 bucks, worth it for me just to cut to the chase.
And I apologize to the pro's out there that know what they are doing, as I am just a trial/error kinda guy.
February 9, 2006 @12:25am

I just joined this site and not quite sure how it functions, so hopefully the right folks see this post.
Like many here, I'm going nuts trying to get effective hiss reduction with no negative effects on the desired sound capture. The format I work with makes it especially challenging: old analog concert audience recordings [ARs]. My present tools are Cool Edit Pro [2002] and Ray Gun Pro [latest]. I've had some success with them but remain frustrated by not enough hiss reduction and varying degrees of destruction of desired sound. By definition I'm looking for more hiss reduction with zero side effect. I realize there's no perfect technology for this, at least not yet, but would be grateful for any recommendations of products anyone thinks superior to what I have for the need. I fully expect a better product to cost something. Thanks for any and all replies! Please email if possible in addition - time limits my ability to navigate this forum extensively
August 13, 2006 @08:48pm

(1) Some programs and/or processes are better than others.
(2) Results vary considerably depending upon the values given to various NR parameters.
(3) A good noise sample provides better results than a lesser noise sample.
Both Hiss Reduction and Noise Reduction are available in CoolEdit/Audition . Almost everyone reports better removal of tape hiss with NR than with the Hiss Reduction transform. Those who have nay-sayed this have never, as far as I know, provide any details.
I also use the Sonic Foundry NR plugin. I find it especially better for one major aspect of hiss removal. I have seen comments, but no evidence, that the latest edition of Audition NR is improved in this respect. This aspect is:
When NR is applied it does not effect the audio file evenly. It is more effective where the real audio signal is lower. The result of this uneven response is a noticeable increase in non-removed hiss just before the audio signal level increases and a delayed falling off of non-removed hiss after the audio signal level decreases. This is often called "pumping." NR can be done so this is essentially unnoticeable, but often only at the expense of leaving more overall hiss than is desired.
Other than this "pumping" problem, NR can add strange noises and/or remove valuable audio. The adding problem can be eliminated by proper NR use. The removal of wanted audio is always a matter of compromise if very heavy duty NR is needed, but some amount of useful NR is always possible.
Noise reduction use starts with an adequate noise sample. This means a selection of recording with the noise you want to remove and none of the audio you want to keep. Whether or not that exists in your recording is perhaps a matter of good or bad fortune after the fact, but noise-only is something you can often plan on capturing if you make the recording.
Listen to your noise selection at extra high volume to be sure you hear none of the audio you want to keep.
Sometimes it is possible to get a useful, if less than perfect, noise sample by filtering out what isn't the noise. For instance, apply a high pass filter to remove valuable audio but leave in the higher frequency noise/hiss; then make the noise profile. This particular process is done by first copying the noise sample to a new file and manipulating it there rather than applying destructive filters to your recording.
A good noise profile in CoolEdit/Audition will require at least 2.5 seconds of noise containing no proper audio. If only shorter intervals exist in your recording, copy and/or duplicate short sections into a new file from which to make your noise profile. (It can be best to adjust the boundaries of your sample to zero crossings before doing the copy step so there are no clicks between the multiple pastes in the new noise-only file.)
The following values for making the noise profile are to assure high resolution, which means a better chance of effecting the noise but not the audio. The noise selection must be long enough to support the high number of snapshots or the program won't make a profile.
Snapshots in profile: (12,000 to) 24,000
FFT: 24,000 — you may want to read in the Help about FFT size and potential problems with time smear. I find this generally not a problem but situations may vary. In general, values suggested for parameters in the CE help files are often much smaller than optimum.
Then apply the noise profile to the audio:
Noise Reduction Level: 100
Reduce by: 1 to 40 dB
Precision Factor: 9
Smoothing Amount: 1 (to 2)
Transition Width: -10dB
Optimum results may require variations in all of these parameters but they are a good place to start. I rarely need variations with any parameter except‘Reduce by.' You can read about each parameter through the Noise Reduction Help button. Again note my comment about the Help suggested values often not being optimum. Variations in the Transition Width parameter may be most relevant to the pumping problem.
Just a note. Identical results (as far as I can determine) are available by either following the above or by setting ‘Reduce by' to a constant value and varying the ‘Noise Reduction Level' slider (or doing some combination of the two). I've settled on only varying ‘Reduce by' as being simpler and no less effective.
CoolEdit NR has a preview function but it isn't always adequate. Select a few seconds or so in your recording, adjusted boundaries to zero crossings, apply NR to just that selection, then listen from before to after the selection. Use the undo button to remove the NR. If necessary, go back into the dialogue to adjust the ‘Reduce by' value until you achieve the best results. This is slower and clumsier than Preview, but often more satisfactory. Or, just do this for a final confirmation of your chosen parameter values.
Another useful test is to copy the audio selection to a new file. Copy it again so there are two copies that will play one right after the other. Apply NR to one of them. Click on the "Play Looped" button to listen to the two variations as often as necessary to decide if the difference made by NR is completely desirable.
This DX plug-in works from within CoolEdit/Audition and probably within any other DX compatible editor.
Sonic Foundry will make a profile on a very short duration noise selection but the profile will be different, and better, with a selection of 6 to 8 seconds.
Reduce noise by: This parameter get the most variation. Try 10dB to start.
Noise bias: 0dB
Attack speed: 100
Release speed: 100
FFT size: 16,384
Overlap: 75%
Reduction type: mode 3
Fit size: 2,048 points
Sonic Foundry, unlike CoolEdit/Audition, does not make a separate noise profile for each channel. Therefore it can be beneficial to do each channel separately when there are significant differences between them.
There are claims that multiple passes of NR, each doing a lesser amount of reduction, will give superior results. I think this is so to a small amount but my tests have not demonstrated very significant differences. To utilize this approach, Apply NR, then make a new noise profile from the processed file, then apply NR again, etc.
Signal on tape is low level, so the tape deck preamp must have quite a bit of gain. This introduces some noise from the preamp and that is very similar to tape hiss. Some tape deck preamps are noisier than others, thus if it is possible to do the transfer from a superior deck, you end up with a lower noise level.
A mixer or line level preamp is sometime necessary between the tape deck and the soundcard in order to get a good input signal level. Some tape decks have a variable level line out. With these two in tandem it may be possible to optimize gain staging. For instance, I have a tape deck with variable level output. Noise falls off rapidly as I dial back the output level from 100% to 70% (measured as rotation of the dial). Reducing the level below 70% has very little effect upon noise. Thus I get best results by setting the output at 70% and amplifying in the next stage.
Dolby NR should be on in the tape deck whenever the tape was recorded with Dolby NR. This is much better than any software NR. I do mostly voice-only stuff from tape. I've found that Dolby NR is very useful regardless of whether or not it was used when recording. In fact I now use Dolby C for all these transfers.
I have not had any music tapes that needed much NR so I have not tried adding Dolby when it isn't part of the original recording. Using Dolby will change the frequency balance, but that may be better than the hiss in some cases. If I had a mucic recording with significant hiss, I certainly would not hesitate to make multiple short recordings of the same tape selection to find out if the results of applying Dolby C (or B) were more desirable than the tape hiss.
Many tape recordings do not have much high frequency audio. Even if the ultimate results are for CD, where 44,100kHz is required, a lower sample rate may give better results by eliminating the higher frequencies that are nothing more than noise. I now record from the voice-only tapes at 22,050 for that reason.
It may be desirable to do the transfer recording at 44,100 because that is too time consuming to have to do over, but once on computer the recording can be resampled to 22,050 (or even 16,000kHz) or filtered sharply with CoolEdit's FFT filter. It is certainly worth experimenting on some short selections to find out what the results are like.
August 14, 2006 @05:51am

Wow - thank you Andy! That's quite a volume of info and it should be very useful to me as both CEP and Ray Gun Pro's manuals are strangely vague on the subject.
I should have mentioned initially that I have been using isolated noise floor samples in CEP for some time with reasonably good results - the current project came to me fully transferred with no isolated tape noise sections, after needed EQing its hiss became fierce and I have no choice but to use blind passes with no samples. That's all Ray Gun does and its been more effective than CEP for such, but results remain unsatisfactory.
Even when noise floor samples are available as they have been when I received discs with entire unedited tape sides, I still have much to learn about nuances of FFT size, drag points [which points to drag and where], decay rate and so on. The manual provides almost no data here.
Again I thank you for the extensive post and theory it contains. I will study it carefully and apply all I can to the next session. I'm not yet set up for analog to digital transfer btw, so far just receive everything already transferred for me. Your advice here is gratefully received nonetheless as eventually I intend to start transferring tapes myself.
Your comments on the Sonic Foundry [Sound Forge? Sony?] plugin are most intriguing. Can this be purchased separately from what appears to be their larger suite ['Acid'] and downloaded?
I did some research and found two outfits claiming superior, 'cutting edge' noise reduction software, one put out by an outfit called Coyote-something and the other called Cedar. I sent them queries but have no replies yet. Anyone here know anything about them?
Meantime if you're curious I can email a FLAC of a short sample of what I'm working on presently. I'd love to hear what Sonic Foundry might be capable of. Email is reem@mindspring.com.
All the best!
August 14, 2006 @07:29pm

Cedar has a long established reputation. Their flagship product(s) is hardware that works in real time between the source and the soundcard. I would guess software is something newer to compete in the current market. They have always been priced very far above my means, so I have not paid much attention.
The Sonic Foundry NR plugin (now owned by Sony I believe) is sold separately. As far as I know, it is not part of any other program. It definitely is not the more primitive NR stuff in their Sound Forge editor. It is a download but I don't recall, from several years ago, if it is at all functional before you pay for it.
If you can get your client to record a little blank tape of the sort the audio was captured on, you can probably use it to make functional noise profiles. It is less than ideal because the tape recorder and playback decks both contribute to what you want to remove, but better than nothing.
Better yet, if this client wants good results, why can't he/she be persuaded to just include some leading and trailing end stuff in what they send to you. It is for their benefit, after all.
August 14, 2006 @10:52pm

i prefer multiband window-gated downward expansion, but i've been coding my own plugs for almost 10 years... so i'm not sure what's available commercially.
i've played with Voxengo Redunoise some too, and it seems to do a pretty good job, fairly clean. http://www.voxengo.com/product/redunoise/ I havn't used a Voxengo plug when appropriate that didn't have the quality I need, and very affordable. recommended. :)
September 14, 2006 @09:14pm