Continuing the Quest to Preserve Four-Track Audio
In 2023, Reformatting Specialist Leif Johnson took on a particularly tricky audio preservation challenge: figuring out how to capture recordings made on four-track cassettes. The blog post about Leif’s work explored the unusual format and the challenges involved in figuring out how to preserve the recordings.
Now, Leif’s project is completed! Leif recently installed a new four-track cassette workstation at the Utah State Archives and developed a detailed workflow for capturing and preserving these recordings.
Among the recordings waiting for this workflow are four-track audiocassettes from the Board of Pardons records. Preserving these recordings requires more than simply playing the tapes and saving the sound as a digital file. Each tape may require adjustments to its playback speed, sample rate, and channel configuration before it can be preserved as an archival file.
From a Challenge to a Workflow
Four-track cassette recordings present several challenges for digital preservation. Unlike a standard stereo cassette that contains two independent audio channels to create multi-directional sound, these recordings contain four independent audio tracks that need to be captured and preserved separately.
The new setup uses a Tascam PortaStudio 464 and a Focusrite Scarlett 4i4 to route each of the four tracks independently into a computer. From there, Leif uses Adobe Audition to capture, process, and export the audio.
The equipment provides the foundation for the process, but the digitization itself requires several additional steps. Each of the four tracks must be captured separately and preserved as an independent audio stream.
Four Tracks, Four Separate Streams
A four-track cassette can contain four independent audio streams recorded onto the tape. Preserving those streams means keeping them separate throughout the digitization process rather than combining them into a conventional stereo recording.
During capture, each track is routed to its own input and recorded as an individual track in Adobe Audition audio editing software.
The hardware is configured so that the four outputs from the PortaStudio 464 connect to the four inputs on the Scarlett 4i4. Each input corresponds to one of the cassette’s four tracks, allowing all four streams to be captured independently.
Capturing all four tracks separately is only the first part of the process. Before the entire tape can be captured, Leif also needs to determine how the recording was originally made, including the speed at which it was recorded.
Finding the Right Speed
One of the first challenges is determining the original recording speed.
The PortaStudio plays tapes at 1⅞ inches per second (IPS), but the audio on a cassette may have been recorded at a slower speed to fit more recording time onto the tape. Playing a recording back at 1⅞ inches per second when it was originally recorded more slowly would make voices sound too fast and too high-pitched.
Leif begins by capturing a small portion of the recording and testing standard speed adjustments in Adobe Audition. His workflow accounts for several possible original speeds, including 1¼ IPS, 15/16 IPS, and 15/32 IPS, with each requiring a different amount of stretching to return the recording to its original speed.
If the recording does not sound correct at one of the standard speeds, Leif has another method for determining the answer through its electromagnetic network frequency.
When the Tape Gives You a Clue
The recording itself can provide a clue through an unexpected source: electrical hum.
In the US electrical power grid, electrical network frequency (ENF) oscillates at 60 Hz. That frequency also produces harmonics at multiples of 60 Hz, such as 120 Hz and 180 Hz. Because the court logging cassette recorders were connected to electrical power, the recordings can contain traces of this frequency.
Leif can examine these frequencies using a spectrogram, which provides a visual representation of the frequencies present in an audio recording. By analyzing the hum and its harmonics, he can calculate how much the recording needs to be stretched to return it to its original speed.
In one example from Leif’s documentation, a frequency analysis produced a peak at 188 Hz. The nearest 60 Hz harmonic was 120 Hz. Dividing 188 by 120 produces a stretch factor of 1.567, or 156.7 percent. That adjustment produced the correct playback speed for the recording.
Sometimes the first method does not provide a clear answer. In those cases, Leif can examine the spectrogram more closely and look for frequency bands that correspond mathematically.
One example showed frequencies at 618 Hz, 207 Hz, and 181 Hz. Because 618 Hz is approximately three times 207 Hz, Leif determined that 618 Hz was likely the third harmonic of the 207 Hz hum. That meant the 207 Hz frequency was likely the original 60 Hz electrical hum, allowing him to calculate a 345 percent stretch factor.
The process is a good example of the detective work involved in audiovisual preservation. When the original recording speed is not known, information embedded in the recording itself can provide the evidence needed to reconstruct it.
Getting the Sample Rate Right
Determining the correct speed is not the end of the calculations. Leif also has to account for the speed adjustment when choosing the sample rate used to capture the tape.
The goal is to produce an archival file at 48 kHz. If a recording is captured at 48 kHz and then stretched by 200 percent, the resulting effective sample rate would be only 24 kHz. To avoid that loss, Leif captures recordings at a higher sample rate when a larger speed correction is required.
His workflow uses different capture rates depending on the stretch factor. A recording requiring a 200 percent stretch, for example, is captured at 96 kHz so that the resulting file can be preserved at 48 kHz. A 156.7 percent stretch, like the example above, is closest to the 150 percent category, so Leif captures it at 88.2 kHz.
Once the appropriate speed and sample rate have been determined, the entire tape can be captured.
The four tracks are recorded, adjusted to the correct speed, and exported together as a four-channel, 48 kHz, 24-bit WAV file. The archival WAV retains the four audio streams as separate channels. Metadata is also added to explain that the file contains four independent streams from a four-track courtroom cassette and that an audio editor is required to access all four streams.
Building a Repeatable Process
Solving the technical challenges of one recording is useful. Creating a process that can be repeated for many recordings is even more valuable.
Leif created five Adobe Audition multitrack templates for 48, 88.2, 96, 176.4, and 192 kHz capture. The templates provide a consistent starting point for future digitization projects.
He also documented the hardware setup, software configuration, speed calculations, capture process, and export settings. The documentation provides a detailed workflow that can be followed when the equipment is used for future projects.
The result is more than a working piece of equipment. Leif has established a repeatable process for capturing a format that requires specialized equipment, careful analysis, and several technical adjustments before a preservation file can be created.
Preserving audiovisual records often means finding ways to connect older technologies with modern preservation systems. Leif’s work has established that path for the Archives’ four-track cassette recordings, making it possible to preserve the separate audio streams and carry their contents forward into the digital archive. That process will now support the preservation of the Board of Pardons recordings and other four-track audio records in the Archives’ collections.
Legal Disclaimer:
EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.
