search

Preservation of Audiotape & the Dolby Noise Reduction System

A Introduction for Assisting Identification, Playback, and Transfer
by Joshua Ranger

The Noisy Business of Recording Audio

Making an audio recording is essentially the process of capturing sound waves on some sort of carrier (disk, tape, etc.). This requires some sort of device to funnel and amplify the sound waves (an “ear” like a microphone) and a device to etch or record those sound waves onto the carrier.

Early recording devices were very mechanical in this process – scratching a physical instantiation of the sound wave onto wax, metal, etc. – creating its own noise in addition to the sound of the recording in the process. This method was paired with amplification devices that were trying to pick up as much of the sound wave’s variations in tone and volume as possible. Aiming for this wide range lessened focused capture control, thus other sound in the vicinity was captured along with the main signal.

The change to more electrical processes and to the use of audiotape reduced the mechanical noise, but recording devices still contained moving parts and points of friction (the tape running across posts or the recording heads). The sound of this movement and friction picked up by the amplification device and recorded on the carrier often results in the hissing noise we commonly associate with audiotape recordings.

There is thus a distinction in the sound present on a single audio recording: program noise (the original music, voice, etc.) and surface noise (the sound created as a result of the recording process).

Listen to this clip of a song recorded impromptu on a tape recorder in a hallway for the hissing and also what sounds like the remnants of an audio signal formerly on the tape.


Bob Dylan – Forever Young

Of course, extraneous noise can also be a part of the program noise. In the next clip you can hear a dog barking and some whining feedback towards the end.


Bob Dylan – Every Grain Of Sand

If this were being produced for an album, the artist would have to record it over (after putting away the dog and fixing the audio levels). In the case of the first sample, a re-recording would present the same surface noise. The only way around this would be a process that lowers the amount of audible surface noise without significantly distorting the program noise.

Dolby Noise Reduction System

Surface noise was an inescapable fact of recorded audio until around the 1950s when there was a push towards discovering a practical noise reduction technology. There are other sources for a more technical explanation of the science and technology behind noise reduction. Here we will keep the discussion to two basic concepts: noise modulation and masking.

Masking is essentially the notion that lesser frequencies are not discernable in the presence of a louder / larger frequency range – the lesser frequency is masked or hidden by the larger one. Think of watching television – the audio from the program is what your ear picks up, not the electric humming of the TV, the fridge, the air-conditioning, the traffic outside… Mute the television, and suddenly those noises become apparent.

The most basic concept behind noise reduction is that the program noise can be used to mask surface noise through modulation. In a song, loud passages naturally mask surface noise, so the problem becomes how to mask the quiet or solo sections where the program noise is not great enough.

If the song were recorded at a high volume, the quiet sections would play louder and mask the surface noise. That, however, would result in a distortion of the the loud sections by overloading the system. Making the recording level quieter during those louder sections could solve this, but upon playback, the sections meant to be quiet would then be loud and the loud sections quiet. The solution to this would be to manually adjust the volume during playback to compensate: Turn it down during the quiet sections and turn it up during the loud sections.

Besides the impracticality of this kind of manual noise reduction, it can also create distortions in the program noise if the recording or playback volume is adjusted too quickly or set at too high/low of a level. You may notice the sudden change in volume, or it may cut out or add in the wrong frequencies if not adjusted correctly to correlate between the frequency range of the program noise and that of the surface noise. Like peeling an apple, you can cut too little and leave bits of the skin on or cut too deep and lose the meat of the fruit. The right kind of knife for doing the work makes all the difference.

In the mid-1960s, Ray Dolby developed an electronic method of automatically performing noise modulation in order to utilize the effects of masking. His system processed the audio signal during recording and then reverse processed it during playback in a way that significantly reduced the noticeable amount of surface noise on audiotape recordings.

Over the years, Dolby Labs developed several versions of the Noise Reduction System, none of which are compatible with each other. To correctly playback a Dolby processed tape requires a deck that has the right Dolby processors engaged at the correct settings. The correct equipment and settings play an important role in the playback, transfer, and preservation of audiotape, which will be addressed in more detail below. First, we will explore some examples of Dolby and non-Dolby recordings.

Dolby NR on Commercial Recordings

Listen to this clip of a commercial audio recording produced without any Dolby Noise Reduction:


Van Halen – You Really Got Me Now (No Dolby)

Note how, when the guitar is playing solo, there is a distinct hissing noise accompanying it – the non-program surface noise resulting from the recording process. You may notice that the noise is not so apparent once the rest of the band kicks in. This is the masking property discussed above. A louder program noise is covering the frequency of the surface noise, thus this would be a section where noise reduction would be de-emphasized during recording.

Relate this surface noise to the “extra” program noise audible: the carry through or resonance of the guitar, cymbals, and vocal exclamation (a slow fade of the sound created, not a clipped stop to it), or the chunkchunk sound of the alternating quick strokes on the guitar.

Though Dolby Noise Reduction quickly became an industry standard for the recording of audiotape, it is important to remember that NR was not used on every single recording. In the example above, the left in “noise” could be considered an artistic choice. There is a rawness and “simplicity” to the sound that makes a statement about the band’s image as loud and stripped down rock ‘n’ roll in opposition to the more produced sounds of 70s super groups and disco. The extra noise is all part of the sensory assault. Hearing the percussiveness of the fingers on the guitar strings as well as every bit of reverberation also allows one to feel the skill and physical reality of the guitar playing.

Now listen to this sample of a commercial recording produced using Dolby B Noise Reduction:


Toots and the Maytals – Get Up, Stand Up (Dolby B)

There is a noticeable reduction in surface noise compared to the previous recording. Also note how the program noise is much more evenly layered and the instruments cleaner sounding – the snare drum presents more muted vibrations and there is less carry through in the guitars.

This of course is partly a stylistic difference. The clipped guitar notes of reggae are achieved through a different approach to playing and for a different effect than fuzzy, elided rock guitar. However, production techniques such as noise reduction can accentuate these variations in performance technique. Preservation of audio settings is not just about technical fidelity, but also about maintaining artistic choices.

Dolby NR on Field Recording

As mentioned above, Dolby Labs developed different versions of their noise reduction system, both at the professional level (Dolby A, Dolby SR) and consumer level (Dolby B, Dolby C, Dolby S). The systems apply the same basic principles, differing in the complexity of how those principles are applied and at what frequency range. For example, Dolby B reduces the noise on a tape recording by about 9 decibels. Surface noise – and any program noise – that falls within that range of 9 decibels louder than the Dolby standard is effectively eliminated.

Listen to this spoken word example which repeats the same short text three times, first recorded without Dolby, then with Dolby B applied, and then with Dolby C applied. Note how the hissing surface noise we are so used to with non-professional recordings is cut out, but also how the audio quality becomes less bright and perhaps more muffled sounding.


Definition of Phonograph from The Devil’s Dictionary (No Dolby, Dolby B, Dolby C)

Dolby B is a simple system that applies mostly to a “white noise” level. Dolby C is more complex. It reduces noise by about 15 decibels. This covers a greater range of surface noise but, when applied to a tape that is processed for a range of 9 dB, cuts out that extra 6 dB of the audio signal. Some of this 6 dB may be surface noise, but some of it may be program noise as well. When this is applied to a tape originally processed without any NR, one would effectively be cutting out 15 dB from the higher and lower ends of the audio signal. Depending on the original volume of the audio, this could create significant distortion or dropout from the recording, clipping off the resonance of the instruments and vocals or causing an effect like someone moving closer and further away from a microphone while speaking.

Playback and Transfer

Depending on your generation, you may have grown up with a standard radio/cassette player with a switch that said “Dolby NR On/Off”. And one may have played cassettes on that stereo, switching back and forth between on/off, yet not really knowing the difference, or perhaps thinking, “Hmmm. Off is louder. Loud is good. I will keep it switched off.”

However, the NR settings are as important as any other setting to the audio, aesthetic, and historic quality of the recording. One would not listen to a 33 1/3 rpm album at 78 rpm nor watch a 3:4 film/video stretched to fit a widescreen television. The original recording choices matter, and the way a recording is transferred and played back matter.

Leaving ethics aside for the moment, what matters in the misapplication of NR settings is the addition or subtraction of segments of the audio frequency in regards to the original signal. The Dolby Systems are digital in concept – they apply averages and algorithms to an analog signal to simplify the relay of complex information. The system hedges its bets, assuming that the selected range will cover most standard deviations. Where it doesn’t quite cover, it should be close enough to not be noticeable.

However, the system can overcompensate and end up greatly distorting the source material, like when a digitized image becomes pixilated if its size is increased too much. The correct NR settings will reduce surface noise and perhaps other extra-program noise. The wrong settings will begin to decrease noise it shouldn’t or will add noise that wasn’t present in the source. Listen to this example of a calibration tone where the settings were switched from No Dolby to Dolby B and then Dolby C as an example. Each section last about 7 seconds:


Calibration Tone

If we look at the sound wave at those transitions points, we can see how the height of the wave decreases as the higher and lower ends (the “noise”) are cut out by the noise reduction process:


Transition from No Dolby to Dolby B

Transition from Dolby B to Dolby C

Sound Produced with No Dolby

Listen to the examples below of transfers made with the Dolby NR set correctly and then incorrectly:


Van Halen – You Really Got Me Now (transferred with no Dolby)


Van Halen – You Really Got Me Now (transferred with Dolby B)


Van Halen – You Really Got Me Now (transferred with Dolby C)

Notice that the hissing sound is not present in the B and C setting. However, this “clean” sound detracts from the atmospherics of the original – it seems less full and present. Another symptom of the incorrect settings is the way the instruments sound increasingly cut off as a greater frequency range is de-emphasized by the subsequent Dolby versions. This is like when a guitar player holds the strings down after strumming in order to immediately stop the reverberations and sound. In the original we can hear that, rather, the extended, overlapping carry through of the instruments and vocals is a part of the aggressive noise aesthetic of the recording.


Wayno – La Pampa y La Puna (transferred with no Dolby)


Wayno – La Pampa y La Puna (transferred with Dolby B)


Wayno – La Pampa y La Puna (transferred with Dolby C)

As with the other sample, the surface noise hissing is de-emphasized in the B & C versions. In this case it is not a stylistic additive, and the Dolby B version might seem superior on first listen. Upon closer examination, problems arise. The sound is definitely more muted – the higher end of the notes are erased, resulting in less contrast among the layers and weakened attack on the notes. There is also a slight breakdown in the audio – the wind and “water” have some frequencies missing and there is some drop out in the guitar line. These same issues are more apparent in the Dolby C version: The water is barely audible, the beginning of the wind is missing, and the guitar line modulates in and out, sounding empty and pixilated. Listen back and forth among the samples to hear how the available frequency range varies.

Source Produced with Dolby B

Listen to the examples below of transfers made with the Dolby NR set correctly and then incorrectly:


Toots and the Maytals – Freedom Train (transferred with Dolby B)


Toots and the Maytals – Freedom Train (transferred with Dolby C)


Toots and the Maytals – Freedom Train (transferred with no Dolby)

The breakdown from the correct setting to the incorrect is quite apparent. In the Dolby C version we can hear the rhythm guitar sound clipped, and there is some wavering and crackling throughout. In the no Dolby version, that wavering, crackling, and dropout becomes more pronounced. Notice too, when the vocals come in, how the high end of the sound is more emphasized. Things sound bright or tinny, and there seems to be separation between the layers of vocals/instruments without the lower frequencies to pull the sound together.


Lucious Brown – Fall Into the High (transferred with Dolby B)


Lucious Brown – Fall Into the High (transferred with Dolby C)


Lucious Brown – Fall Into the High (transferred with no Dolby)

This sample displays how difficult it can be to tell the difference between correct and incorrect settings. There seems to be little variation between Dolby B and Dolby C, and the no Dolby transfer seems to add only the surface noise and a brighter tone to the music which cannot be immediately discounted as the wrong setting. Luckily the source tape is marked as Dolby B so we know the correct setting. But what about non-commercial recordings, the kinds of items that are more commonly found in archives and less commonly marked with the correct settings?

Unmarked Sources and Field Recordings

The following examples are from a mix tape and a field recording – tapes which would most likely be unmarked regarding noise reduction or other settings. Both happen to have been recorded with no Dolby NR, but listen to transfers of them made with the Dolby NR set correctly and then incorrectly:


Sinéad O’Connor and the Chieftains – Foggy Dew (transferred with no Dolby)


Sinéad O’Connor and the Chieftains – Foggy Dew (transferred with Dolby B)


Sinéad O’Connor and the Chieftains – Foggy Dew (transferred with Dolby C)

As in earlier examples, you can hear how the overall sound becomes more muted and both the instruments and vocals become more clipped at the ends of notes or lines as the different levels of NR are applied. The greater the frequency range the NR system is designed to mask, the more “noise” (i.e., frequencies within that range, including quieter and higher pitched audio like breaths and reverberation that seem similar to hiss or white noise) is removed by the Dolby processors. And even though the recording, being a 10+ year old mix tape, is not the most pristine, there is still obviously a more correct transfer. The audio quality of the original recording needs to be overlooked when considering fidelity to the original signal for determining the correct archival transfer. Though, as always, transfers for access or production may consider other factors of audio quality.


El Gulfo – Intro (transferred with no Dolby)


El Gulfo – Intro (transferred with Dolby B)


El Gulfo – Intro (transferred with Dolby C)

This is an example where a better access or production copy might be improved with a little adjustment. The transfer without NR applied is the correct setting, but there is a lot of surface noise and a brightness to it that does not add much to the recording. The transfers with NR applied are slightly muted (though only sounding off in the audience reactions), and they seem to create or accentuate a channel shifting (the audio emphasis moving between left and right) which may be less noticeable in the non-Dolby version, but the decrease in surface noise makes them more pleasant to listen to. Both preservation of the original signal and creation of a more produced version have their own value and uses which are up to the preservationist to determine, such as the case of creating an unmastered Preservation Master and a mastered access or distribution copy. There is a great artistry to archiving and preservation. It should not be feared, but must be applied with training, practice, and care.

Take a Listen for Yourself

Mixed Sources

Listen to the examples below of transfers made with the Dolby NR set correctly and then incorrectly. Take a close listen and identify the audio clues to variations among the transfers. Which would you determine to be correct? Do you think you can identify which system is which? Answers are below.

Sinéad O’Connor and the Chieftains – Foggy Dew


1. Sinéad O’Connor and the Chieftains – Foggy Dew

2. Sinéad O’Connor and the Chieftains – Foggy Dew

3. Sinéad O’Connor and the Chieftains – Foggy Dew

Toots and the Maytals – Get Up, Stand Up


1. Toots and the Maytals – Get Up, Stand Up

2. Toots and the Maytals – Get Up, Stand Up

3. Toots and the Maytals – Get Up, Stand Up

Handyman Awards


1. Handyman Awards

2. Handyman Awards

3. Handyman Awards

FunKadelic – Can You Get to That


1. FunKadelic – Can You Get to That

2. FunKadelic – Can You Get to That

3. FunKadelic – Can You Get to That

Breaking Up Is Hard to Do


1. Breaking Up is Hard to Do

2. Breaking Up is Hard to Do

3. Breaking Up is Hard to Do

Answers

Sinéad O’Connor and the Chieftains – Foggy Dew (Mix Tape: No Dolby)

1. Transferred with Dolby B
2. Transferred with Dolby C
3. Transferred with no Dolby

Toots and the Maytals – Get Up, Stand Up (Commercial Recording: Dolby B)

1. Transferred with Dolby B
2. Transferred with no Dolby
3. Transferred with Dolby C

Handyman Awards (Field Recording: No Dolby)

1. Transferred with Dolby C
2. Transferred with Dolby B
3. Transferred with no Dolby

FunKadelic – Can You Get to That (Mix Tape: No Dolby)

1. Transferred with no Dolby
2. Transferred with Dolby B
3. Transferred with Dolby C

Breaking Up Is Hard to Do (Commercial Recording: No Dolby)

1. Transferred with Dolby C
2. Transferred with no Dolby
3. Transferred with Dolby B