Using filter settings to correct 'Dolby B"

[charlesC]

Has anyone any experience in using some filter settings to correct (remove) Dolby B recording emphasis?

I have a number of music concerts etc. that were recorded on to 10.5 inch tapes on a Revox B77 some 20 years ago, using a Dolby B unit. The Dolby unit no longer works, but I would like to recover the recordings.

I looked through the other postings, but could not see anything obvious.

Would appreciate some guidelines. Thanks.

[Dan McDonald]

Re: Using filter settings to correct \'Dolby B\"

Charles -

You might want to see if there's a website that tells you what Dolby does (or Craig can probably say), and then use DCart to reverse the process.

I have had good luck producing good sound by simply using the paragraphic equalizer to emphasize the parts that need expansion. I don't know if that's the right way to do it or not.

Dan

[Craig Maier]

Re: Using filter settings to correct \'Dolby B\"

The correct filter capable of reversing Dolby B is the Dynamic Noise Filter. I am not sure of the perfect settings, but I might consider setting the frequency at around 2.5 KHz and the gain at -6 dB in non-enhancement mode. Then adjust the threshold level til you find the sweet spot where the high end is more balanced and not shrill. I do not know what attack and release time constants that I would consider. I suspect 50 mSec attack perhaps, and maybe 200 mSec release? Anyone know the correct numbers?

[Dan McDonald]

Re: Using filter settings to correct \'Dolby B\"

The Dolby website (www.dolby.com) lists some of the technical specs. If you click on Technical Information, then on Noise Reduction, you get an article that describes the differences between Dolby B, C and S and how they work. I hope that helps.

Dan

[OpAmpMan]

Re: Using filter settings to correct \'Dolby B\"

The Howard W. Sams "New Audio Cyclopedia" has a good discussion including graphs showing encode and decode curves with frequencies and levels for Dolby "B" systems. See pages 907 - 908, I believe.

[charlesC]

Re: Using filter settings to correct \'Dolby B\"

Thanks for everyone's input on this. I did try the low pass filter, and that wasn't bad, but the Dynamic Noise Filter that Craig suggested is obviously the better one to use.
Dan's use of the Graphic Equaliser would also no doubt of helped the emphasised high end (like the Low Pass filter I initially used), but reading up on the Dolby B characteristics, something that triggers at a certain threshold is clearly a better way to go.
I tried it with Craig's initial suggestions, and it sounds better than the Low Pass filter. The attack and delay and other fine-tunings are a lot more difficult (at least for me in a quick test) to really hear the difference, but I will try and do some more detailed tests.
In the meantime I am already enjoying some old recordings made more accessible by being brought out of storage and 'loaded' on the Home Entertainment PC as MP3 files.
By the way, on this point, I was a RealJukeBox user for a long time, until I found Media Jukebox. What a great program, far more extensible (formats, remotes etc). Do check it out, if you haven't already.

[greg_m]

Re: Using filter settings to correct \'Dolby B\"

Dolby B was designed to reduce only higher-frequency tape hiss. I don't have the numbers in front of me, but basically a Dolby B did this:

Below some "midband frequency" (which was probably somewhere between 500 Hz and 1500 Hz... but I'm guessing), everything was recorded flat.

Signals above that frequency, IF they were of a fairly high level, were also recorded flat, and played back flat (in other words, no change in gain in either the record or playback chain).

But if the signals above the "midband frequency" were of a rather low level, then the recording box boosted the HF gain (getting the signal somewhat higher, relative to the tape hiss), and the playback box lowered the HF gain (dropping the signals back to the original level, and dropping the tape hiss, too).

Now as I recall, the amount of boost/cut would be drawn as a sloped line... flat from 0 Hz to the "midband frequency," then sloping up (or down) as the frequency increased above that point... and I think it flattened out again at some higher frequency. The slope of this line would be continuously changing, depending on the frequency content of the program material!

What you really want, then, is some kind of dynamically variable filter, that can be continuously varied between the "maximum HF gain" (which is flat) down to some "lower HF gain" with is the maximum amount of NR.

Hopefully the reference provided by OpAmpMan will give you the correct numbers for the frequency curves and time constants.

One word of warning, though, Dolby A and B are very level-sensitive. When you set up a Dolby tape system, you always needed to play a calibrated "Dolby reference level" tape, and adjust playback gain to produce a certain amount of processing at that calibration level. Unless you can come up with such a test tape, and calibrate your playback system, you will have a tough time making your DSP processing accurately track the original Dolby processing on the recorder.

In fact, Dolby's level sensitivity was one reason that dBx systems gained such popularity; they were NOT level sensitive as the Dolby system was. Dolby was great in a studio environment, where there were accurate meters on every tape deck, and the decks were aligned and adjusted frequently... lots of trimpots, lots of maintenance engineers, etc. But when Dolby got into the home arena, with various makes of decks, and countless different tape oxides (each with its own sensitivity and frequency response curve!!!) then it became a real mess. In fact, the only thing worse than Dolby B on home reel-to-reel was Dolby B on cassettes (ugh!).

Good luck!

[greg_m]

Re: Using filter settings to correct \'Dolby B\"

more on topic...

I started looking at Dolby's Website, and found a lot of hype (er, information) about Dolby B. Even some useful frequency response and gain curves. (When I reached the part where they say a Dolby S cassette sounds better than a CD, I decided I had read enough.)


I think four issues remain:


1. Dolby describes Type B as using a filter with both sliding gain and sliding frequency breakpoint. At some gain settings, there even appears to be a little bit of a peak in the frequency response.
So Q: can anything in DC-Art duplicate that sliding frequency filter? I've read the description of the Dynamic Noise Filter and it's not clear whether it's fixed or sliding frequency. I've recently communicated with Curtis and Craig about this filter, maybe I will understand it better, soon.


Q2: The Dolby stuff I've read so far doesn't talk about time constants (it's "consumer sales literature" rather than engineering specs). In addition to duplicating the frequency curves, one needs to duplicate the time constants, too, so that the (software) decoding will correctly track the (hardware) encoding. Where the heck do we get this info? Or do we need to find a working unit and run some analog tests to see what it does?


3. Dolby B was level sensitive (unlike dBx, which was not, but that's another story). You calibrated the tape machine to put out a given voltage at the tape's magnetic reference level (originally I think this was 185 nano-Webers/meter, but may have changed later to 250 nW/m). Then you calibrated the Dolby unit relative to the tape playback. Their Figure 4 shows that gain is unity from 100% down to some threshold point, below that it changes dyamically.

So to make a playback/restoration "system" that really tracks Dolby B, we need a tape deck that's reasonably flat, AND a Dolby level calibration tape, so we can adjust gain within the digital domain to the Dolby calibration point. I wonder if this is reasonable or even possible at this point in time... hmmm...

4. Have the patents expired, or will Dolby sue us if we succeed?

[Dan McDonald]

Re: Using filter settings to correct \'Dolby B\"

Maybe it would be easier if someone with an open-reel machine with Dolby B offered to transcribe the reels to cd for Charles.

Dan

[greg_m]

Re: Using filter settings to correct \'Dolby B\"

No doubt, Dan! I'd offer to do it myself, if I had a Dolby B; but I don't.

(By a strange twist of fate, I have a Dolby card cage with seven "type A" cards, but they are four-band units, whereas "type B" is single band; I don't know of any way to make the "A" cards correctly track a "B" tape.)

Any volunteers care to step forward?

[Bob]

Re: Using filter settings to correct \'Dolby B\"

Check ebay. There are numerous Dolby "B" units for sale cheap. An Advent or Dolby-brand unit will sound much better than any attempt to duplicate the real thing. There were other makers of Dolby "B" processors, but so far I like the Advents best of the consumer units.
When playing Dolbyized tapes, you can, of course, just roll off the high frequencies --which was what Dolby recommended when playing back Dolby-processed tapes not using real Dolby playback -- but at low levels when those frequencies are exaggerated, it's going to sound somewhat artificial; at high levels, Dolby "B" is not really engaged at all, so anything you've done will adversely affect the frequency response during louder passages.
Good luck!

[greg_m]

Re: Using filter settings to correct \'Dolby B\"

I tend to agree with you, Bob, that using an actual Dolby B unit will be a more accurate playback than trying to reproduce the Dolby curves and time constants using generic software (i.e. software that was not specifically designed to do that).

I didn't remember that Dolby had licensed other companies to build standalone Dolby B units, that's intereesting.

Perhaps Charles (who is in France, BTW) will manage to locate one for use with the tapes he has.

I didn't remember Dolby's advising to play back with highs turned down; as you say that would perhaps let the lower level HF's sound *approximately* right; but would roll off the higher level HF's.

I do recall that one of their claims was that if playing back on a "marginal" system (e.g. car tape player, cheap boombox) without Dolby B, the only net result would be that the lower level HF's would be boosted slightly, which would counteract road noise, or poor frequency response of the boombox, etc.

At any rate, good tip about EBay, thanks!

[Beautman]

Re: Using filter settings to correct \'Dolby B\"

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by greg_m:

I didn't remember Dolby's advising to play back with highs turned down; as you say that would perhaps let the lower level HF's sound *approximately* right; but would roll off the higher level HF's. <HR></BLOCKQUOTE>

I don't know if that was Dolby's advice. It was frequently printed on prerecorded Dolby B cassette inserts.

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by greg_m:
[B]I do recall that one of their claims was that if playing back on a "marginal" system (e.g. car tape player, cheap boombox) without Dolby B, the only net result would be that the lower level HF's would be boosted slightly, which would counteract road noise, or poor frequency response of the boombox, etc.

[B]<HR></BLOCKQUOTE>

It also helped out the highs on lower quality tape stock, as long as you backed off the record levels a little to avoid high end saturation distortion.

[greg_m]

Re: Using filter settings to correct \'Dolby B\"

Right-ho.

By the way, there are two on EBay this week, one in the US, and one in the UK...

Charles, there's the quick solution to your restoration project!

[Craig Maier]

My measurements on a Concord DBA-9 Dolby B Noise Reduction unit yields the following Time Constants:

Attack: 20 Milliseconds

Release: 40 Milliseconds