Play It Again, Paul vs. Brand X (September 2019)
In all my years of converting audio media, I had never expected the opportunity to arise where Play It Again, Paul could go head-to-head with a competitor. Yet such a situation arose. While I was engaged in a project converting magnetic tapes, my customer hesitantly inquired about converting some records. The story unfolded that another company had digitized those records onto CDs but, in her opinion, had delivered an unacceptable product. I took the records and borrowed the CDs to ascertain whether her disappointment stemmed from the quality of the audio or from inflated expectations. What I discovered confirmed the former, and the records were added to my current project.
What follows is a comparison of what I did with the records and what the first company appears to have done as inferred from audio and physical evidence. I will borrow an overused advertising approach from the 1960s that compared the sponsored product with “Brand X.”
The Media
There were four nitrocellulose lacquer records (see Uncommon Media), heavily played, and of special emotional value to the customer:
-
One 10” record from 1940, 78 rpm, one track on each side, the longer lasting 3:34;
-
One 12” record of a 1951 concert, 33 1/3 rpm, the longer side lasting 11:31; and
-
Two 12” records of a 1955 concert, 33 1/3 rpm, the longest side lasting 11:28.
For the records from 1951 and 1955, it appeared to me that the concerts were taped, and the records were subsequently cut onto lacquer blanks as the tape was played back. When a song didn’t fit in the space remaining on a side, the tape was stopped when the record ran out of space, backstepped, and restarted with a bit of overlap after the record was changed to a blank side. There were three such occurrences where a song was started on one side and finished on another.
Conversion
Play it Again, Paul: Though the records appeared “full,” i.e., almost all the available surfaces had grooves cut into them, the longest side lasted less than 12 minutes. Since vinyl LPs were designed to hold approximately 25 minutes per side, I concluded that these lacquer records were made with a larger cutting stylus, which is consistent with the history of lacquer blanks being used for 78s. I therefore used a 78 stylus.
Brand X: The amount of surface noise remaining in the CDs’ tracks suggests that Brand X used a stereo stylus, which would have had less contact with the grooves’ sides where sound vibrations were cut and would have picked up noise from the bottom of the groove.
Equalization
Play it Again, Paul: Since the mid-1950s, RIAA pre-emphasis equalization that adds volume in the treble and reduces volume in the bass is applied before masters are cut to overcome some technical problems (see Equalization). Modern preamps do the opposite by applying de-emphasis (i.e., removing the pre-emphasis) when a record is played, thereby restoring the original sound qualities. With this project, however, the earlier two records predated adoption of the RIAA convention. Moreover, direct-to-disc cutting machines used for lacquer records lacked the electronics to apply RIAA pre-emphasis equalization. Consequently, the likely outcome of the conversion process would be tracks with exaggerated bass and diminished treble. I confirmed this for all four records by observing a spectrum analyzer display while listening to the audio. I therefore applied equalization to undo the preamp’s RIAA de-emphasis, which restored the audio to what it sounded like originally: reduced bass, increased treble, and a more natural sound.
Additionally, since lacquer records made from a direct-to-disc cutting machine were incapable of capturing high frequencies, I subsequently applied equalization to reduce frequencies above 10 KHz, as they would have been nothing but noise. Figure 1 displays a spectrum analysis of the result from one of the tracks.
Brand X: The audio uniformly sounded muddier (high bass) and muted (thin treble). This is reflected in spectrum analyses such as the one appearing in Figure 2, where the volume of low frequencies is substantially higher compared to the level in Figure 1. What appears to be a higher volume in the treble in Figure 2 was likely due to additional noise in the treble range. The treble in the musical instruments definitely sounded quieter in the Brand X CDs. At any rate, it’s apparent that Brand X did not remove RIAA de-emphasis. You will also note that frequencies above 10 KHz (pure noise) remain in Figure 2.
Figure 1: Spectral analysis of Grieg’s Piano Concerto in A Minor—Play It Again, Paul version.
Figure 2: Spectral analysis of Grieg’s Piano Concerto in A Minor—Brand X version.
Noise Reduction
The processes employed by Play It Again, Paul and Brand X to reduce tick, click, crackle, and broadband noise produced similar results. However, it’s likely that Brand X’s use of a stereo stylus captured more noise and less signal, and some of that noise survived their noise reduction process.
Waveform Editing
Play it Again, Paul: I spliced the three songs that began on one side of a record and finished on another into single tracks. Fortunately, there was overlap of the last part of the first fragment and the beginning part of the latter fragment, and the redundant audio could be removed. (The customer was unable to identify where the splices occurred.) Extensive editing was performed to remove pops and other noises that “survived” noise reduction software.
Brand X: Brand X identified one of the three songs that were split between sides and successfully combined the fragments into a single song. The other two remained as they appeared on the record. Numerous pops remained in the finished tracks.
CD Burning
Play it Again, Paul: Since several of the sides contained one complete song and a fragment of a second, I split complete songs into separate audio files, which allowed for proper titling of the tracks. With the customer’s concurrence, I burned the final versions of the files from all four records onto a single CD. I then tested it on two sound systems to ensure audio quality and presence of the CD-Text information (see Labeling Music CDs: CD-Text).
Brand X: Tracks from the 1940, 1951, and 1955 performances were burned onto three separate CDs. At least one track burned onto a CD appeared to be a version preceding noise reduction. The two sides of one record were burned out of sequence, resulting in the first track containing the second part of “Song A,” and the second track containing “Song B” followed by the first part of “Song A.” (It should be noted that the records themselves were very poorly labeled.)
Metadata and Labeling
Play it Again, Paul: Performance dates were found on the records’ labels, and I obtained track titles, artists’ names, and an agreed upon CD title from the customer. This information was incorporated into the CD-Text file burned onto the CD and WAV files for all tracks, the latter provided to the customer on a flash drive for back-up purposes in case the CD was later damaged. The artists and CD title appeared on the front of the CD case insert; the track titles, artists, and performance dates appeared inside the front cover. The CD was labeled by hand with the CD title and artists’ names.
Brand X: No CD-Text files were burned onto the three CDs, and there were no inserts. Default track names (Track 1, Track 2, etc.) incorporating performance dates were printed on the CDs along with the company logo, but there were neither CD titles nor artist information.
Cost
Both companies digitized the original media, applied noise reduction software, and provided CDs in slim cases. Despite differences in pricing structures—base price plus options vs. fixed standard price, the cost to the customer was very close. Play It Again, Paul charged $1.65 more, for which the customer received the benefits of an appropriate stylus, post-conversion equalization, a CD-Text file, and a jewel case insert.
Play It Again, Paul performed extensive waveform editing to remove pops and other spurious noises. Though this increased the cost, the customer approved the initial cost estimate and periodic revised projections as the project progressed and was very satisfied with the ultimate benefits for that cost.
Conclusion
I would characterize the Brand X business approach as performing their services rapidly and in a cost-effective fashion. Beyond conversion and noise reduction, add-ons such as inserts and CD-Text files are considered to provide little of value, and the more time-consuming work of adjusting processes to uncommon media, tailored equalization, and waveform editing is not offered. This approach should work well when customers’ media are commercially produced recordings like LPs and cassettes, but not so well for home recordings or very limited productions such as these records.
What Play It Again, Paul does is very similar to what Brand X does given commercial recordings (except that I consider inserts and CD-Text files to be important). But when it comes to home recordings or limited productions, applying the same workflow as Brand X is unlikely to yield the highest quality conversion, and I personally would be embarrassed providing so-so deliverables. Since I’m often being entrusted with priceless, one-of-a-kind recordings, I want to do whatever I can to produce the best possible audio. That usually entails investigation and experimentation, and if I don’t like what I’ve done, I’m prepared to discard everything and start over (and have done so many times). And it invariably entails a lot of communication with the customer throughout the project. I receive my greatest rewards when customers are delighted with what I’ve done, as was the case with the project described above.
Back to Paul’s Blog & Contents