Welcome Guest ( Log In | Register )

How to Clean up Cassette to Digital, Need Help
post May 3 2003, 00:27
Post #1

Group: Members
Posts: 9
Joined: 22-April 03
Member No.: 6123

Thanks in Advance for any help one of you might offer.

I am far from an Audio Pro, yet have been trying to take some old cassette tapes and restore the sound quality they originally had, and enhance it even. As for this file in particular it seems to have lost its high's and mids. When I run it through a simple EQ plugin I just dont get the clarity I need, it just sounds worse.

Here is the steps I took in recording this tape.

#1 - Used a brand new Sony Tape deck , with preamp mixer and got the sound quality as close to 0 DB as I could (as reccomended by many audio recording sites I have read). I recorded at 44.1 hz, 16 bit through my SBLIVE card. I saw no loss in quality from tape to digital.

#2 - I then loaded Wavlab and watched the spectrum analyser and saw the highs and some mids were very low. I tried to raise this up in value but it just made the sound even more distorted than it already was.

I have many tapes like this I would like to restore, re-master (or whatever the term is). I have attached this file (in original .wav format) and uploaded it to the alt.binaries.sound.mp3 newsgroup. The topic is called:

"Sound Pro's can you give tips on how to fix this Audio File? Email - Trick@lvcm.com"

You can also download the .wav file clip at - http://www.lvcm.com/trick/badwav/badwav.wav

Any help on cleanup (what software/settings to use), examples , web links on fixing this sorta of stuff is GREATLY appreciated!!!! Heck if someone wants to clean it up themselves and email it back with the steps you used to do it, that would even be better.

This advice would be Pricessless, thanks again in advance to anyone who can help this audio newbie.

Email - trick@lvcm.com



PS - I have read about the CEDAR system, is this hardware based only???
Go to the top of the page
+Quote Post
Start new topic
post May 7 2003, 13:25
Post #2

Group: Members
Posts: 265
Joined: 12-January 03
Member No.: 4542

First, I'd agree that you should leave some headroom. I've left at least 2 dB at the peak of the loudest track when pushing it close but being careful by pre-auditioning, and 5 to 6 dB is safer in case you find an unexpected peak. Usually tape hiss will be louder than noise from your soundcard inputs and ADC (assuming you mute/deselect noisy unused inputs in your Volume Control panel Properties/Recording), so there's no point risking clipping to improve signal-to-noise ratio by 0.1 dB. Even a cheapo "PnP 16v" soundcard I used has about -60 dBFS input noise floor, which is better than a tape with Dolby.

Yes, it's usually type B for commercial recordings on cassette tapes.

This is roughly how Dolby B NR works:

The frequencies above about 10 kHz can undergo heavy dynamic compression without introducing audible harmonic distortion (because the harmonics are above 20kHz and inaudible, and are filtered out anyhow). So Dolby boosts the quieter high-freq sounds to higher levels when recording the tape, while reducing those that are already loud by a smaller degree so that they are mostly well above the level of the hiss that's inherent in the tape and mostly at >10 kHz (partly because of the speed of cassette tapes being low to extend playing time on a compact cassette).

When you play back on ordinary equipment, the treble sounds particularly bright, but hiss isn't reduced. Because the quiet sounds are made louder, cymbals and hi-hats can often sound extended in their decay and the timbre (harmonic overtones) of some instruments is overly bright compared to the original. To some, subjective quality is high (they like this treble-boosted sound), but in fact fidelity is low. The fact average people liked this modified sound allowed Dolby-B mastered cassettes to be sold by default by record companies. Even without Dolby, you could reduce hiss and maintain a pleasant sound by turning down the treble slightly.

Turning on Dolby-B applies dynamic range expansion that is the inverse of the compression applied during recording, again, only applying it to those high frequencies. This brings boosted low-level signals back to their correct low levels and at the same time, reduces tape hiss (which starts at low level) down to a much lower level, reducing this source of unwanted noise. Any harmonic distortion introduced at this time is above 20 kHz so is filtered out and inaudible anyway. Now, the cymbals and hi-hats will sound less extended in their decay, much as they ought to sound, and the timbre of instruments will be more like the original sound and quiet passages will have much less high-frequency hiss. Regardless of noise, it's the high-fidelity of transient decays and recovery of the correct musical timbre that leads me to use Dolby-B NR if the source was recorded that way.

Additionally, Dolby made demands on the acoustic quality and specs of equipment that bore their logo, so it usually assures a decent frequency response and reasonably low noise electronics, for example.

When I was into cassettes I was a student on a tight budget, and my best audio cassette player remains a Sony WM-36 Walkman with Dolby B, not part of a separates system. With the volume below maximum, it has low distortion, and with the EQ all set at 0 dB it has pretty low noise - probably about -55 dBFS on ferric mastered tapes. I have recordings with apparently legitimate 16 to 17 kHz frequencies present at the sort of places one would expect (transients, cymbals etc) and have made some pretty decent-sounding restorations. (PM me for a link to a website to listen to one before and after and see the technical details of restoration and software used). That deck's big failing is it won't play some cassettes with high-friction mechanisms at a constant speed, so in some cases I'd have to break them open and put them in a good cassette housing (which I've never tried yet) to use that tape deck.

Like Pio, I'm also cautious of digital noise reduction, having tried it in CoolEdit96 some years ago (mostly with broadband FM receiver noise, which is trickier than tape hiss when it's at a high level) and noticing some artifacts both in boundary-effects (FFT smoothing/windowing) and in tinkle. However, this may have a lot to do with CE96's implementation and something to do with the loudness and character of the noise and the amount of reduction I wanted to apply.

I've recently tried Exact Audio Copy - Process WAV - Noise Reduction. This has a 0 to 48 dB scale of reduction, not a knife-edge setting of 0 to 100% or more than 100%, which seems to respect the statistics of gaussian noise better and be easier to set to a not-too-aggressive level. I found it worked remarkably well with about 24 dB reduction with fairly critical listening. 24 dB was enough for me, so I didn't push any further.

I think I'll try to set up a blind test on this forum soon so that people can compare different software and settings and ABX them effectively.

For the blind test I'd supply a few minutes of noise from a commercial tape with long lead-out, sampled at 44.1 kHz, 16 bit on a fairly cheap soundcard of about 1999 vintage "PnP 16v". I'd also supply a separate 3 second sample from the lead-in of the same tape from which to feed the noise reduction routine with a sample of noise-only (tape hiss remaining after Dolby B NR + soundcard noise). Testers could then mix-paste the noise with CD music of their choice (preferably with <100% peaks to avoid clipping) and try to ABX the difference between original WAV and the WAV that had the hiss added then Noise Reduction applied. We could then try to determine which kinds of sound are most degraded by these routines, and which settings and software packages are considered safe, if any.

Perhaps we can also suggest improvements (e.g. apply less NR if the noise is psychoacoustically masked by the music), though I suspect that it will always be difficult to avoid degradation during fade-outs as the signal approaches the noise floor when viewed in the frequency-domain (it can theoretically fall below the apparent noise floor on waveform view and still stick far above the noise on the spectral view). I tend to think that noiselike signals (broad spectrum, like cymbals, snares, hi-hats, vocal sibilants), rather than continuous tonal signals, may be the first to get damaged as the volume decreases towards the noise floor. I have some interesting jazz samples with brushed snares that could be demanding tests.

Anyway, any of you can feel free to send me a PM if you want a link to a website to hear and read about my recent restoration which includes 'before' and 'after' samples of EAC's noise reduction on a very quiet lead-in to a quiet track only ever released on cassette. I'd welcome criticism from any of you. I had to compensate for background noise when auditioning the settings by using quiet clips and boosting the signal, so I'm not completely happy to assume I'd never notice a probem.
Go to the top of the page
+Quote Post

Posts in this topic
- cupelix   How to Clean up Cassette to Digital   May 3 2003, 00:27
- - dreamliner77   JUst a thought If you are recording the wav's...   May 3 2003, 00:41
- - TwoJ   Hi Cupelix PSS - Cedar is a company that does aud...   May 3 2003, 01:49
- - Pio2001   24/96 soundcards should bring nothing for recordin...   May 3 2003, 02:17
- - cupelix   Thanks for the helpfuls replies guys. In response ...   May 3 2003, 02:49
- - TwoJ   Sorry Pio2001 - I didn't mean to imply that sw...   May 3 2003, 03:18
- - woody_woodward   This topic is fascinating to me. I've done se...   May 3 2003, 06:19
- - Andavari   I'm interested in what the results would and w...   May 3 2003, 06:48
- - JeanLuc   QUOTE (Pio2001 @ May 3 2003 - 01:17 AM)1-I sh...   May 3 2003, 09:37
- - Pio2001   QUOTE (cupelix @ May 3 2003 - 04:49 AM)As far...   May 3 2003, 14:39
- - cupelix   I dont know if any of you bothered to download the...   May 4 2003, 00:36
- - woody_woodward   I tried downloading the WAV from your original pos...   May 6 2003, 19:15
- - Xenion   Nakamichi Tapedeck + Nakamichi High Comp II => ...   May 6 2003, 19:21
- - Mac   I want to have a go with your file, but I also rec...   May 6 2003, 22:09
- - _Shorty   usually when it doesn't state which type of Do...   May 7 2003, 02:56
- - DickD   First, I'd agree that you should leave some he...   May 7 2003, 13:25
- - JeanLuc   QUOTE (Pio2001 @ May 3 2003 - 01:39 PM)An MA-...   May 7 2003, 14:43
- - 2Bdecided   I think the same name has been applied to differen...   May 7 2003, 17:19
|- - knutinh   QUOTE (2Bdecided @ May 7 2003, 17:19) IIR...   Oct 17 2013, 07:56
|- - 2Bdecided   QUOTE (knutinh @ Oct 17 2013, 07:56) QUOT...   Oct 17 2013, 12:47
- - gleib   QUOTE (Pio2001 @ May 3 2003 - 05:39 AM)QUOTE ...   May 7 2003, 18:48
- - raynault   I do A LOT of Tapes, mine are all voice. First u...   Oct 16 2013, 05:26
|- - pdq   QUOTE (raynault @ Oct 16 2013, 00:26) Fir...   Oct 16 2013, 13:33
- - 2Bdecided   As neither tapes decks nor soundcards universally ...   Oct 16 2013, 14:20
- - AndyH-ha   There is nothing in the signal path to vary the le...   Oct 16 2013, 22:01
- - raynault   Usually the "pre amp" in the tape player...   Oct 24 2013, 17:40
- - 2Bdecided   QUOTE (raynault @ Oct 24 2013, 17:40) Dav...   Oct 25 2013, 09:54

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:


RSS Lo-Fi Version Time is now: 17th April 2014 - 00:31