Friday, March 24, 2017

Behold The RIAA Curve!

The RIAA has a bad reputation. But despite all of the horrible, unethical, hypocritical, soulless, malicious and devious things that RIAA has done... they still deserve credit for the RIAA curve. Prior to 1954, recorded audio had a problem. A world of wholly un-standardized playback and recording  equipment used across independent recording studios, radio studios etc had led to a wild west of audio quality. Differences between needles, reproducers, cartridges, microphones, and recording media made for mixed audio characteristics everywhere. Today the RIAA equalization specification permits longer playback times, and better, more consistent sound quality.  More here.

The RIAA 500R-13.7 equalization curve has operated as a de facto global industry standard for the recording and playback of vinyl records for over half a century. Before 1954 (especially from 1940) each record company applied its own equalization, each pressing plant, and recording studio... it has been estimated that there were over 100 combinations of turnover and roll-off frequencies in use.  (Audacity software actually has a EQ library you can install to emulate the numerous differing curves.) Below are just some of the more common standards.
  • Columbia-78
  • Decca-U.S.
  • London LP
  • BBC 2dB
  • Bartok 629C
  • NBC Ortacoustic
  • RCA Orthacoustic
  • Columbia LP
  • Decca-FFRR-78
  • AES
Each EQ curve is a combination of two filter characteristics: a bass boost curve which is defined by a Bass Turnover frequency, and a treble cut curve, defined by a Gain Rolloff parameter. Early 78s, especially in the US had no curve. (Not on purpose anyway). The EQ is flat and to the modern ear it sounds like it. This is unsurprising when you recall that the speed wasn't even consistent between record labels leading to utterly unavoidable differences in pitch. This was exacerbated by excessive patent hoarding wherein each company felt what they were doing used proprietary technology they often wouldn't license.

The earliest EQ standard I know if is the European 78, which dates to at least 1926. It had a pronounced boost in the bass range at 50 Hz. Nearly as old is the Blumlein 300, which has no boost in the high frequencies and a turnover at 300 Hz. It's named for Alan Blumlein a genius electrical and audio engineer at EMI. (He may have also invented stereo sound.)  But these (and many others) had developed for different purposes. The NAB standard was for radio transcriptions. RCA developed a standard for it's "Orthophonic" recordings in 1947 only to change it in 1952 to find a happy medium to pair with their new 45 rpm recordings.  More here.

The NAB standard was probably key to the later standards. The lateral cut NAB curve was remarkably similar to the NBC Orthacoustic curve. They were both boosting low frequencies not for mathematical reasons but to over equipment hum. When the Columbia LP was released in June 1948, it's EQ drew heavily on the former and used a bit more bass boost. The took the unusual step of publishing the information. AES followed suit in 1951 with the explicit goal of standardizing home stereo systems. But RCA and Columbia were adhering to dissimilar standards and it went nowhere.

Then in 1952 RCA published their EQ standard, with an article by R.C. Moyer claiming a technical lineage back to the Western Electric "rubber line" recorder in 1925. This argument was clearly intended to establish theirs as a "historical" standard. That Western Electric Rubber line was a magnetic disc cutter to which we can attribute the birth of EQ. It was a well engineered cutter, and maintained a constant velocity. This allowed engineers to detect a correlation to amplitude and frequency. So with the goal of avoiding amplitude distortion bass was boosted after recording the first deliberate EQ manipulation. This is credited to Joseph P. Maxwell and Henry C. Harrison of Bell Labs.

Then in July of 1954, the Engineering Committee of the RIAA published the RCA standard attempting to standardize all their member labels. Electronic Industries and Tele-tech magazine published a rebuttal claiming it was the same as the NARTB standard, an update of the NAB curve. This was not exactly true. But in reality the Columbia, RCA and RIAA curves were all similar by this time [What's a couple kHz between friends?] If anything the RIAA closed the gap between the other standards, a political decision as much as an engineering decision, but all three standards (RCA, Columbia, and NARTB) were totally workable. It was more important at that time to pick one and move forward.