Restoration Tips & Notes » recording/mastering

Capturing both directions of a half-track mono tape

Richard L. Hess — Fri, 06 Jan 2012 20:39:50 +0000

It is possible to capture both directions of a two-sided half-track mono tape in one pass.

The critical factors are:

Azimuth
Direction
Polarity

The azimuth of both sides needs to be the same. If the two sides were recorded at different sessions and/or on different machines, then there is no guarantee that azimuth will be the same on both sides. In a large-volume project, this can be addressed by installing two repro heads, one for each direction, and adjusting azimuth separately. This does not work well for stereo tapes with different azimuths because any temporal perturbations in either the recording or reproduce passes will result in severe stereo image shift. If a record head had azimuth scatter between the two channels, it is best to split the difference, if possible without severe high-frequency attenuation.

Obviously, the direction of the playback needs to be reversed in the DAW (Digital Audio Workstation) software, but that happens pretty quickly on a current computer.

Not-so-obviously, the polarity of the signal is also reversed when played backwards. That, too, can be addressed, possibly in the same pass or, worst case, a separate pass in the DAW software.

This is not recommended for any tapes using noise reduction such as Dolby or dbx, nor for the highest quality of music. However, most two-track tapes recorded on both sides are not of this quality.

Zoom H2 line input

Richard L. Hess — Tue, 14 Apr 2009 04:23:23 +0000

The Zoom H2 HandyCorder is perhaps the lowest-cost digital recorder on the market that provides reasonable and useful results. While I have a Sound Devices 722 for my more serious work, I bought the Zoom to test it out to see if it could be part of a simple tape digitization system for archives on a budget who wish to do the work themselves. It does this reasonably well.

As with much equipment–and especially with lower-cost equipment–the performance specifications and the actual operational data is not published. There are reports of the H2 clipping on the line inputs in some of the reviews and it appears that a lack of understanding how the inputs were configured exacerbated that situation.

There is nothing wrong with the line inputs on the H2. BUT there are some caveats:

DO NOT use the input level control on the line inputs to go below 100 or the preamps will clip before the signal reaches 0 dBFS.
ALTERNATE WORDING (thanks Greg H.):
Set Zoom H2 RECORD LEVEL to 100 or greater to avoid clipping at the Line In preamp stage.
Use an external attenuator with the gain set at 100 to avoid overdriving the line inputs.
The noise floor is not spectacular, but is not too bad. With the inputs terminated in 150 ohms, the peak noise was -70 dBFS, but that improved to about -85 dBFS or better, measuring it as an A-weighted rms figure, which is how most noise is measured. While this certainly isn’t what one would expect out of the Sound Devices, it is far better than the 50-60 dB(A) that one can achieve with an analog cassette tape without Dolby.
The maximum input level to the line input should be no more than -5 dBV or -3 dBu.
Try to avoid clipping as there appears to be a delayed recovery in some instances.

With this information, you can optimize a pad between the source and the H2 line input so that the recorder is never overloaded. Try to keep the levels as high as practical as there is a relatively limited dynamic range. On the other hand, I have found that the recorder noise is not objectionable even when I’ve boosted the levels 20 dB for a quiet choir piece from our church. The room ambience totally swamps the recorder noise–at least as far as I hear.

The H2′s internal mics are reasonably good for many purposes. While I still prefer the SD722 for many things, I think the H2 is one of the better oral history conversation capturing devices I’ve seen. It uses SDHC cards which may require a new card reader. Don’t use the built-in USB connection unless you’re recording MP3 files as you’ll be there all day. The internal card-reader mode is limited by the USB 1.1 interconnection. This interconnection is adequate, however, for using the H2 as a 16-bit live audio interface/microphone connected to your computer (at either 44.1 or 48 ks/s). The unit will record up to 96,000 samples per second, 24 bits, but many of those bits will be noise.

Please let me know if this has helped or if you have any questions or comments.This work was done with version 1.50 of the H2 firmware.

StoryCorps experience including equipment discussion

Richard L. Hess — Mon, 04 Feb 2008 02:13:31 +0000

My friend Susan Kitchens and her brother took their parents to the StoryCorps recording session in Los Angeles a few weeks ago. She blogged it here. One of the neat things is that between her article and the discussion she and I had in the comments, we have a good handle on most of the equipment that was used in the trailer. It’s a good selection in my opinion and shows how simply good-quality recording systems can be set up. Further discussions from a StoryCorps representative have shown how clever the setup is.

Tape recorder bias frequencies

Richard L. Hess — Sat, 02 Feb 2008 20:44:02 +0000

The discussion of what bias frequencies were used over time keeps recurring. Special thanks to Jay McKnight of Magnetic Reference Lab, Tom Fine, and Brian Roth for input to this list. I posted this to the ARSC list, but wanted to include it here as well. This knowlege is useful for those who wish to archive the bias along with the audio for future application of time-base-error correction tools such as the Plangent Processes.

Bias frequences started low, apparently with 60 kHz for early consumer recorders, but Ampex started with 100 kHz. Other later machines used different bias and erase frequencies. We can see with a few exceptions, the top bias frequencies were commonly limited to 250 kHz for audio, with the Sony APR series and the Ampex ATR series in the 400 kHz region. For cassettes, a practical maximum appears to be about 150 kHz. Much higher frequencies (up to at least 8 MHz) were used in instrumentation recorders. An enumeration of several machines follows.

Wire and Reel Tape Audio Recorders

In the early days, apparently wire recorders used bias as low as 30-40 kc, but Jay McKnight recalled in the pre-Ampex days, 60 kHz was common.

Ampex

The Ampex 200A used a bias frequency of 60 kHz [Jay McKnight 2012-03-25 post to Studer List] He indicated that this was probably the lowest for any professional recorder.

The Ampex Standard was 100 kc up to the MR-70.

With the MR-70, Ampex switched to 150 kHz bias frequency (and adopted the Hz) [Larry Miller, ex Ampex]

Ampex AG-440 (A) stayed with 150 kHz [manual]

Ampex ATR-100 144 kHz erase, 432 kHz bias (1:3) [manual]

MCI-Sony

MCI JH-24 Multitrack 210 kHz bias, 105 kHz erase [manual via Brian Roth]

Sony APR-5000, APR-24 100 kHz erase, 400 kHz bias (1:4) [manual]

Otari

Otari MTR-10/12 II Bias 250 kHz (erase not spec’d) [manual]
Otari MTR-90 (original) 246 kHz bias, 123 kHz erase [manual via Brian Roth]

Studer

Studer A80VU (All versions) 80 kHz erase, 240 kHz bias (1:3) [manual]
Studer A80 RC 150 kHz [manual]
Studer A810, A807, A820 2CH, A820 MCH, A827 153.6 kHz [manual]
Studer B67 150 kHz [manual]
ReVox A77 120 kHz [manual]
ReVox B77 150 kHz [manual]
ReVox PR99 150 kHz [manual]

Tascam

Tascam 32/44-OB — 150kHz [manual via Tom Fine]

Technics

Technics 1500/1506/1520 — 120kHz [manual via Tom Fine]

Cassette Audio Recorders

Here is a quick sampling of published bias frequencies for two top-of-the line cassette recorders, a better-than-average portable, and an early compact portable.

Nakamichi Dragon (Along with the Nakamichi CR-7A, perhaps the finest machines ever made for overall audio quality) 105kHz [Service manual dated 1985 (scan) 1990 (copy)]
Nakamichi MR-1 — 105kHz [manual via Tom Fine]

Studer A710 (a high-end cassette recorder, without the auto-azimuth that makes the Dragon superior) 150kHz [no date, scan on Studer ftp site]

Sony TC-D5M (a workhorse, good quality stereo portable) 85 kHz [Svc Manual dated 1980]
Sony TC-55 (an early compact — jacket pocket — mono portable) 41kHz (as low as I’ve ever seen) [Svc Manual dated 1972]

Reel Instrumentation Recorders

Honeywell 101 Medium Band: 4 MHz, Wide Band: 8 MHz [1977 manual] (note that the direct recording bandwidth on Medium Band was 600 kHz and 2 MHz on Wide Band at 120 in/s)

Racal Store / DS (Dual Standard) series 1.2 MHz [1982 manual] (note that the direct recording bandwidth was 300 kHz at at 60 in/s, the same wavelength limit as the Medium Band version of the Honeywell 101)

Aligning a tape recorder

Richard L. Hess — Sat, 02 Feb 2008 19:18:03 +0000

It seems some people new to tape are confused over how to align a tape recorder. This is the abbreviated version.

If you want to record on a tape recorder (and I do not recommend doing that these days as you’re just generating more tapes that will need to be transferred later) the first thing to do is get the playback correct.

CLEAN the machine.
If you haven’t done it in the last year or after a move (depending on the machine), demagnetize the heads and guides (using a strong demagnetizer like the Han-D-Mag).
Get a NEW (or trusted) calibration tape from MRL
The MRL tapes are supplied tails out. Rewind carefully and slowly onto a large-hub reel.
The first tone is a lineup tone, set for 0 on the VU meters of all channels.
If you are compulsive, the first time you do this, check the VU meter calibrations using an external AC voltmeter with wide frequency response. Most professional decks have very flat VU meters, so once you confirm that, you can just use the VU meters for the alignment.
There is a second lineup tone at different levels. If it is one of the -10 dB levels, take the machine out of playback cal and increase the level so that the meter again reads 0 VU.
On the 8 kHz azimuth section align the playback head azimuth (with an oscilloscope or a scope-application in the DAW) for minimum phase shift. Also check in mono sum.
Adjust the EQ trims (Trans-treble on the Studer A810/flashing treble light) for 0 VU.
On the 16 kHz tone, readjust the azimuth for minimum phase difference and maximum amplitude as above. Check in mono sum as well. It will never be perfectly stable.
Low frequency adjustment cannot be accurately accomplished off a test tape due to fringing unless the test tape and the play head track width is matched. However, one can often get close a test tape, but don’t necessarily tune for flat. It’s best to leave this alone if you can. The right channel of quarter-track machines will show more bass than the left as the fringing effect is coming in from both sides. Read the material on the MRL website.
Finally, recalibrate the playback level setting on the last tone. Leave the tape in a played wind on the reel it came on.

This completes the playback adjustment. Now you are ready for record adjustment.

Place a piece of blank tape on the machine (NOT your calibration tape from MRL)!
Record a 700 Hz tone at 0 VU on the meter when monitorin input and adjust the record level calibration for 0 VU when monitoring the output. Do this for all tracks.
Increase the frequency to 10 kHz (and drop it 10 dB at slower speeds, making up the gain in the uncal portion of the playback gain controls).
Decrease the bias level slightly so that you can find the peak. Then increase the bias past the peak until the 10 kHz level off the tape drops by the amount specified for that particular tape. It’s often somewhere around 3 dB. There are other, more precise ways to do this, but this should get you close.
Do a sweep of the high frequencies and adjust the HF record equalization for response closest to the response you got from the test tape. DO NOT try and improve the response from the test tape while adjusting record EQ as that will give you non-standard tapes.
Do a sweep of the low frequencies and then you can better adjust the PLAYBACK LF equalization.
Go back to 700 Hz and adjust for 0 on the VU meters when reading input.
Adjust record level control for 0 on the VU meters when reading output.

That should do it. I generally do a quick check flipping between input and output monitoring and you should hear no difference.

A word about levels. In the old days, I used to record at 185 nWb/m with Dolby A. With more modern tapes, 250 nWb/m will provide adequate headroom in most cases and may reduce the need for noise reduction processing. However, some have complained that 250 nWb/m is too low as it sounds too “digital” (i.e. “clean”). If you want to use tape as an effect, increase the record level to taste.

I really love recording with my Sound Devices 722 or somewhat less so with my MOTU 828 MKII, though there is nothing wrong with the MOTU that an RME FireFace 800 wouldn’t fix! Of course, now MOTU has the new 828 MKIII and it seems they have improved some of the things I complained about, but … twice burned (8Pre, also) … Anyway, I think that quality digital recording will capture sounds closer to the original than analog magnetic tape. This has been true in most tests run since the early days of digital recording and why most of the classical engineers who are looking for accuracy and not colouration were early adopters of digital. If you wish to record on analog that’s wonderful, but consider that analog tape is being used as much as an effect or sound-colourant as it is a storage medium. Also, remember that your legacy of tapes will be much more costly to preserve and migrate than digital files, although they may withstand neglect better.

Doug Pomeroy commented:

After aligning the deck for playback, per Richard\’s list, there is a simple way to set bias on any machine, recording on any tape. Use a 1000 Hz test signal and set bias current for maximum recording sensitivity (VU meter reading). Then for 15 and 30 ips recording, increase the bias until the
output level drops 0.2 dB. For 3.75 and 7.5 ips recording, decrease the bias until sensitivity drops 0.1 to 0.2 dB. One slight problem with this is being able, accurately, to read such small values on a conventional VU meter! (This method comes from Jay McKnight, of MRL Labs.)

Another method, also requiring a tone generator, is to record a low frequency, such as 30 Hz, at a very low level – at least 20 dB below normal operating level – and crank up the playback level enough to hear the output clearly, then adjust bias for the minimum amount of distortion (modulation noise, actually). This method allows one to easily adjust the bias by ear, listening for the cleanest reproduction of the low tone. The point of minimum modulation noise will very closely match bias settings arrived at by more elaborate means.

OF COURSE, after setting bias one must always go back and look at the high frequency response (10 kHz) and readjust the recording eq for flattest response.

Yes, bias setting is somewhat complicated, but it is good to remember it is always a matter of compromise, between the least distortion on the one hand, and the flattest overall high frequency response on the other.

Thanks for the comment, Doug. These are also good ways to set the bias. It is always a tradeoff. I no longer remember all the details, but biasing some old Magnetophonband Typ L from circa 1943 was a real challenge on a modern recorder as it is a homogeneous tape which means the \”magnetic coating\” is much thicker than on any coated tape, so the thickness loss is greater, and its basic sensitivity was far less than even something like 3M/Scotch 111.

Using the proper tools…and don’t try this at home!

Richard L. Hess — Thu, 10 Jan 2008 03:10:08 +0000

I didn’t think I needed to write this post, but it appears that someone purchased a Racal Store 4DS Instrumentation Recorder at least partially because I mentioned it, hoping that it would work as a four-track recorder for creating music.

Well, this is not a good idea. The machine uses non-standard-to-audio equalization. There is no “sel sync” (Ampex TM) to play back previously recorded tracks in time with newly recorded ones. There is no individual-channel erase system (and no erase head at all on the 7DS 1/2-inch, 7-track machine).

While this machine is useful to me for professional tape restoration. With a knowledge of its shortcomings and benefits, it can be a useful tool. I’m afraid, however, that I need to say, “DON’T TRY THIS AT HOME”!

In general, there are things that are posted on this Blog that if used inexpertly can fail to meet expectations and perhaps do damage. In this case, it only caused disappointment.

Specifically, instrumentation tape recorders were designed to gather data in the field and then permit later analysis of that data in the days before digital computers could do this much better. Audio and instrumentation look a lot alike, but are different enough that unless you completely understand the intiricate details, you will be disappointed.

I would also suggest that using the one-piece hard-disk-based digital music-production units from a variety of manufacturers will go much farther to enhancing your creative musical expression than an instrumentation recorder or even a cassette “porta studio”. I am not in a position to recommend any particular unit.

Tape speed, track width, frequency response, and dynamic range

Richard L. Hess — Sun, 26 Aug 2007 19:41:53 +0000

There are complex interactions between tape speed, track width, frequency response, and dynamic range. This article is an attempt to summarize the major influences.

Track Width

Track width primarily influences dynamic range. Since the noise generated by tape is totally random, each time the track width doubles, the noise only goes up by 3 decibels. That same doubling causes the signal to go up by 6 decibels, so each doubling of the track width provides a 3 decibel improvement in dynamic range.

Another factor with track width is that the wider tracks have fewer “micro dropouts” or “grain noise” from the roughness of the tape particles.

Also, wider tracks provide greater protection against dropouts caused by physical damage.

Finally, track width may negatively impact frequency response at lower speeds. As the track width increases, holding high-frequency losses due to azimuth wander to a minimum becomes more difficult. For a discussion of azimuth, please see this article “Azimuth: Hows and Whys” (click here) in this blog. This writer has found that many 7.5 in/s full track 1/4-inch tapes reproduce better with a narrower reproduce head than the track width. Using a full-track head on these tapes results in a periodic high-frequency loss that is very noticeable and cannot be easily corrected.

Speed

Higher speed will provide better high-frequency response within reason. There are two factors relating to this: tape thickness loss and gap loss. Both are important at lower frequencies and gap loss becomes less of an issue at professional speeds (15 and 30 in/s). Dale Manquen has treated this extensively in “Handbook for Sound Engineers”. I will not try to summarize this here.

Higher speed also produces a corollary loss called “head bumps” in the low frequency. This too is related to the ratio of the wavelength to the length of the tape pole pieces and the tape contact with the pole pieces. Generally, 30 in/s low frequency reproduction is inferior to 15 in/s low frequency reproduction.

Higher speeds also average out the irregularities in the tape better than lower speeds.

Higher speeds also tend to have better dynamic ranges, but this is very dependent on the equalization selected. Jay McKnight has discussed this in various papers. I would suggest starting with ” Tape Recording Equalization Fundamentals and 15 in/s Equalizations“.

Azimuth: Hows and Whys

Richard L. Hess — Wed, 27 Sep 2006 16:07:38 +0000

There is a recurring question as to what is the best way to set azimuth for playing a tape. Many people assume that using the test-tape alignment is best. Well, that makes another big assumption: The recorder used a proper test tape alignment. While that can be the case, it usually is not.

If the tape has tones on it (a big if in many collections), you align to that, right? Better, but consider this: When a 1963 album was being remastered, the mastering engineer aligned to the tones and after doing all his magic found that the tones were not recorded at the same azimuth as the program. Big redo!

Aligning to the track material itself is the only way to properly adjust azimuth in the absence of tones and it is still necessary to check the track material even if there are tones. The wider the track and the slower the speed, the more critical the azimuth adjustment is. Stereo makes it more critical and often times easier to adjust.

What is azimuth? It is simply the angle of the recording or reproducing head gap relative to the motion of tape travel. It should be perpendicular. 90 degrees. 90.000 degrees. But there are limits to the accuracy of even the best test tapes. And there are limits to the operator’s ability to adjust the machine. And the machine is subject to vibration and shock in its life. Here is an excellent paper by Jay McKnight of Magnetic Reference Lab about how they do azimuth and the limitations. Figure 6 of this MRL paper shows the frequency response when the azimuth is varied from optimum. What happens is that there is a combing effect and that can not fully be removed by re-equalizing after the transfer. Therefore, despite hardware and software “azimuth correctors” that are available, it is best to align the playback azimuth as best you can prior to transferring the tape — and do it with program.

Properly adjusting azimuth to program takes practice. If you have a mono tape, that’s all you have, but if you have a stereo or multi-track tape, you can sum channels that have similar program on them that should be in phase and get a wider baseline for tweaking the azimuth. Of course, gap scatter in a multi-track head, especially, can corrupt this approach. Gap scatter is the linear distance along the direction of tape motion between gaps in a multi-track head. It should be 0 which means all tracks’ gaps are in perfect vertical alignment. With stereo or multi-track recordings, you can use a phase scope to narrow the line of the mono component. Dale Manquen commented that he thought gap scatter was more of a problem than I made it out to be. He offered the following suggestion:

If you use a dual-trace chopped display triggered off one track, you can simultaneously see azimuth peaking in the amplitude, and time errors in the time offset. If the amplitude and phase adjustments don’t coincide, then you would have a starting point for the time offset due to gap scatter.

Adjusting azimuth is very similar to manually focusing a camera lens. You rock through the point of best azimuth and see how far out it is on either side and try to manually find the centre between two equally bad side points. You tune for MAXIMUM high end. Don’t worry about making a tape screachy or annoying — all you’re doing is capturing what’s on the tape. There is no point of azimuth adjustment that will make the tape brighter than the original recording. Anything but the proper azimuth will produce a tape that is duller than the original recording. The point of proper azimuth is very narrow. Often, the best performance will be within +/- 5 degrees adjustment of the azimuth setting screw. Make sure you’re listening to the highs as you do this. If the cymbals stop, go back to where the cymbals are. If you don’t have cymbals, listen for whatever high end you have. If you don’t have good speakers, use headphones. Remember to listen to stereo programs in mono.

One word of caution, especially with tones: You can set the azimuth to a false peak. This happens most often on narrow tracks. Checking with the voice announcements on a test tape will tell you if you’re close. It’s possible to have 8 kHz in phase on a stereo machine, appearing to be proper azimuth, and still have 4 kHz out of phase. If that happens, go back and tweak the 4 kHz and then proceed up the scale. This error is actually less likely on program material than on tones. The one time this happened to me was the first time I aligned an 8-track 1/4″ head. I wasn’t used to very small difference between the main peak and the peaks on either side of the main peak due to the very narrow (21 mil) track width.

Here is an example of a 1.88 in/s mono two-track tape that was recorded with, I believe, test-tape setup followed by the same section after I adjusted azimuth to the program. In the piano section of the demo, you can actually hear the “combing” effect in the first section. There is some digi-junk even though it’s a 128 kb/s MP3, but you should be able to hear through that.

It is critical to play the tape with the proper azimuth. Remember, at this point in the tape’s life, this might be the last playing as you digitize it. Make sure it’s the best playing you can give it.

Bertrand Navarro commented:

Hello,

I\’ve leave a comment here because i\’ve been working for the past two years on a new system that basically remove the burden of azimtuh adjustment.

I work in a R&D data storage company and we developped a system based on our previous magneto optical data reader.

We demonstrated that with magneto optical reading of the tape we can reach pretty good performances (SNR, Bandwidth, and THD). Plus, instead of using a single head, we\’re generating a laser line across the tape which is then imaged onto a photodiode linear array. The main idea is to reduced the indivual height of each optical head and then lower the constraint on the azimuth alignment. The azimtuh is seen as phase shift between optical signals and can be corrected by simple signal processing. So bassically you just have to set the head close to 0 ° azimuth but this setup can be done with less constraint. Furthermore, is the azimuth changes during playback (Tape written on two different decks or LTM), it can also be corrected by post processing.

I hope this technologies will be push forward to help in the preservation process that is going on.

If you want more information you can contact me at bnavarro@hi-stor.com

Regards

Bertrand NAVARRO

Posted Apr 3, 2007 7:53 AM

How to archive recordings — a quick guide to resources on this site

Richard L. Hess — Wed, 15 Mar 2006 19:40:52 +0000

You’ve been asked to digitize recordings in your collection and don’t have any idea where to start. There are several resources on this site which might be of use.

What I use is shown on my facility page. That\’s one of the main reasons it is there. If I’m using it, it’s because I like it or it solves a problem for me. If I’m not using it, either I don’t have an opinion about it, won’t spring for it, or don’t like it.

It seems the good stand-alone CD recorders (like the Sony CDR-W33) are going out of production. Your best bet is to use a computer. Please consider a good USB/1394 audio interface and a good audio editing program.

Pro Tools is a good program, but I think Samplitude (which I use) offers more bang for the buck and is very, very clean sounding. Many musicians that I know use the Steinberg lineup of software. Adobe Audition (the former Cool Edit) is also widely used.

Browsing all the topics in this section, reviewing the Formats & Resources section, and looking at my facility page should give you a good idea of where to start and why.

I’d be happy to answer any questions submitted via email. Some may be answered here, so please let me know if you wish to remain anonymous or quoted by name/email/website.

Analog "Warming up" of sound & the use of obsolete formats in contemporary recordings

Richard L. Hess — Wed, 15 Mar 2006 04:27:45 +0000

I received a phone call today from someone who wanted my opinion on a Tascam 238 8-track cassette recorder for recording his music.

This was like the person who wanted to know about the DCC recorder for the same purpose yesterday.

People keep hearing that “analog sounds great” or that this or that format “sounds great” and they want to buy in.

I am telling people that if they want to use analog to “warm up” their tracks, it should be a high-quality reel-to-reel machine—or a plug-in. I also tell people when they ask me what to use that perhaps they should look at what I’m using on my facility page. That’s one of the main reasons it is there.

Sure an 8-track cassette (how DO you fit EIGHT tracks into a 0.150-inch tape?) is analog, and it might add warmth to a recording, and noise, and probably mis-tracking dbx if you don’t want the noise, and dropouts, and low headroom.

I have a 238 and it solves problems, the biggest of which is people who used to use this format who want to recover and digitize what’s already on the format.

I also tell people DO NOT use obsolete formats (reels are an exception). Please don’t create more obsolete material that needs to be transferred down the road. This goes double or triple for obsolete digital formats. I would not suggest anyone buying into any of the dedicated digital formats today. That includes ADAT, DTRS, DAT, MD, DCC, or DASH.

There is a trend to look at great older recordings and then try and copy it by getting similar equipment. Well, there are lots of things that went into the older recordings, and, in some instances, the great recordings were made despite the limitations of the equipment. Some old equipment is great, but the marketplace has moved on and the best use the old equipment can be put to is rescuing old recordings made in the old formats — obviously I think that as that is what I’m spending my time doing.

If you have a garage band or are a singer-songwriter, please consider a good USB/1394 audio interface and a good audio editing program. The knee-jerk reaction is “oh Pro Tools.” Pro Tools is a good program, but I think Samplitude (which I use) offers more bang for the buck and is very, very clean sounding. Many musicians that I know use the Steinberg lineup of software.

When is the 20 kHz limitation an issue? How big an issue?

Richard L. Hess — Wed, 08 Mar 2006 05:50:25 +0000

The 20 kHz bandwidth of CD audio media may cause truncation of some material. Here is an example of a small amount of energy above 20 kHz in a symphonic recording. It is interesting to note that this is a 7.5 in/s recording done on 1970s prosumer equipment. I’ve said in my presentations for some years, most 7.5 in/s tapes are well-suited to 44.1ks/s 16 bit transfers, but there are exceptions. This shows one.

Whether or not it’s distortion products or real music has been argued by some. I have been transferring most music programs at 88.2ks/s and 24 bits since about 2003. Here is a recent spectral bar graph that shows a 7.5 in/s tape I made in 1974 played in 2005. This shows the peak spectral energy of a portion of Smetana’s Dances from “The Bartered Bride.” Note the energy in the 25k4 band – this came up with cymbal crashes during a crescendo. This is the original master recording using a pair of AKG C-451 microphones feeding a ReVox A77 on Scotch 207 tape.

Transfer Recommendations

Richard L. Hess — Wed, 08 Mar 2006 05:43:56 +0000

Remember, this transfer that you (or I) are about to undertake may be the last time (and hopefully the best time) that the original is transferred. Here are some suggestions:

Make at least two masters and a listening copy. Keep one set of masters off-site.
Use the best playback equipment you can find. It does make a difference, and world-class equipment is often being sold off at fire-sale prices.
Check playback azimuth. Wrong azimuth problems can only be partially compensated in downstream processing. The bad azimuth artifacts will stay with the recording for the rest of its life if you don’t get it right the first time.
If your tapes have tones, align the playback device(s) to the tones (especially Dolby). Record the tones on the digital copy.
Make sure your noise processing (Dolby, dbx, Telcom) is working properly and is the correct type. There are at least five different versions of Dolby (Pro: A & SR; Consumer: B, C, S) and two of dbx (Pro: Type 1; Consumer: Type 2). There are no software plugins to decode this properly. You must have a properly functioning decoder.
Be creative in non-invasive ways to optimize playback. For example, a cotton swab can be an auxiliary pressure pad during some rough areas of playback on reel machines that don’t normally have pressure pads.
Don’t play a tape with sticky-shed or that appears to be losing it oxide without some treatment.
Make straight transfers before processing. Save these as better noise processing algorithms may be available in the future.
Save a good portion of the noise footprint on the tape without other signal information for later noise reduction processing.
At first, worry less about a final product than getting a good, clean transfer with as few artifacts as possible.
Above all, listen…are you getting the best transfer you think you can?

Comments:

Tom Proctor Posted Mar 14, 2006 2:39 AM

I have been trying to find good information on noise reduction processing using a noise footprint. Does anyone know of a relaible method or filter that can be applied to reduce random noise such as tape hiss? Any suggestions appreciated.

Richard L. Hess Posted Mar 14, 2006 11:36 AM

First, please consider the discussion in this article and generally do not leave the noise-processed copies as the preservation master as this is an area where technology will improve.

With that said, I am a real fan of Algorithmix Noise Free Pro. There are other products to consider: Cedar, Quadriga from Cube-Tec, and probably others. The Dehisser in Samplitude and Algorithmix\’s Sound Laundry are both more limited versions of their big brother. Either can do a good job, it\’s just that NFP does it better and offers more controls.

Tom Proctor Posted Mar 14, 2006 1:25 PM

I am currently trying get information about soundcards. This is primarily in reference to users of PCs that want to transfer recordings with good results but don\’t want to invest in pro equipment. I have had satisfactory results using a 24 bit Soundblaster Audigy 2 ($80), but I see there are a lot of cards out there for a lot more $. Anyone know what makes them better and how much better are they?

Richard L. Hess Posted Mar 14, 2006 1:45 PM

There is an aging resource for this information here. There are, of course, really high-end A-D converters, but there is some consensus about the Digital Audio Labs CardDeluxe as being a good value for money for a PCI-based card.

This is partially discussed here as well and I think that is a better location for ongoing discussions.

Susan Kitchens Posted Mar 27, 2006 9:29 PM

If I may make a small squeak on behalf of MacOS based productsâ€¦ I recently tried out Peak\’s SoundSoap, and it does an okay job. (the demo movie is fantabulous, of course, and your own results vary. But it\’s got a good end-user UI w/o having to go into tons of technicalties) Of course, I do noise reduction on a copy, not the original. Disk storage is cheap.

Archival processing

Richard L. Hess — Tue, 07 Mar 2006 18:32:22 +0000

As promised, I will respond to some questions that are asked via email by answering here in the Blog.

One of the things I’m most concerned with is the appropriate use of digital processing in transcription for cleanup or remastering of digital archival copies. This includes both questions of when (if at all) processing beyond the actual A/D conversion is appropriate, and which are the techniques and currently available tools best suited to archival audio.

It’s a good question. To some extent, it depends on the client and the final use.

If the restoration/preservation reformatting is for an institutional client, then the first transfers should be as unprocessed as possible — at least the initial copies that are archived should be done that way. The main reason for this is that processing algorithms will always get better and they may hide some information that is useful to future researchers–information that today we consider “noise.”

I am conservative when setting audio levels when making transfers because there is no way of knowing the loudest portion of the signal in advance. So I generally transfer at 24 bits and then raise or normalize the level prior to dithering down to 16 bits for the distribution copy. If I’m working on music, I will generally archive the 88,200 ks/s or 96,000 ks/s files before the normalization.

Processing should generally be done on a copy. The exception to this in my mind are private clients who want the best possible copy of their parents’ wedding, or some other important event. If applied conservatively, noise reduction and equalization will be appreciated by these clients and most of them won’t care a bit that it’s been processed. I keep the unprocessed files on my servers until I am sure the client is happy with the processed version.

As to what to use, there are a wide variety of options available. At the high-end, this falls into the category of “remastering” rather than simple restoration and I’m sure there are options that I’m not aware of.

As a first step, I am very pleased with the basic capabilities built into Samplitude. In addition to that, I use Algorithmix Noise Free Pro as well as the Sound Laundry suite. Really tough projects can often be improved by the filters in Diamond Cut 6 Live/Forensics and most of the filters are available in the lower-priced Diamond Cut 6. Diamond Cut and their main dealer, Tracertek, often run sales which was how I upgraded to Live/Forensics.

Other products with excellent reputations are Cedar Cambridge, Quadriga Audio Cube, and many others. Listening to and discussing with other users via one or more of the mailing lists listed here is very useful.