Conceptual Feedback Loops in the Transmission of Recorded Music
In a preceding article, Fidelity in the Transmission of Music, Part I: Three Possible Paths, one important path, Composer → Performer → Medium -→ Listener, was depicted as a sequential or linear sound processing chain. The simplification made was a mechanism to transform multiple signals at source into a perfect representation on the medium released (usually with two channels only) without incurring any loss. This idealised sequence of processes is shown in the following diagram. The emphasis here is mostly on classical music performed and recorded at a venue; the objective in the fidelity of transmission is to create a recording for consumers to listen to at home in order to create an experience as if they were present at that venue.
On that very topic, renowned audio engineer Floyd E Toole states in his book Sound Reproduction — The Acoustics and Psychoacoustics of Loudspeakers and Rooms, “It is naive to think that it is possible to capture a ‘pure’ sound from a musical instrument and to reproduce it without loss”. In fact, the audio signal processing chain involves many stages such as tracking, mixing and mastering, as well as devices, from microphones, equalizers, and an array of analogue and digital signal processing equipment. Most importantly, however, these processing stages are not performed unsupervised or in an automated fashion: skilled recording and mastering engineers influence the parameters of the processing stages and their devices. And to do that, these engineers need some means to make audible what is being produced: amplifiers and loudspeakers. Feedback loops are hence created as shown in the following diagram.
Every time any signal in the chain is “probed”, may it be in the recording control room or in the mixing room, amplifiers and loudspeakers (or headphones) are involved, not to mention people’s hearing and perception abilities. Consequently, what will be heard on the feedback paths will inform the configuration of the signal processing chain, and by extension, what ends up in the medium released can can be heard by a consumer. For instance, if any amplifier or monitor speaker is characterised by a peculiar frequency response such as an amplification of some frequencies, a sound engineer agnostic of this shortcoming will turn some knobs on the mixing desk to compensate that by attenuating those frequencies on the output. A consumer listening to the result on a different system will then perceive these frequencies as being “absent”. Toole quoted, “… it was found that the inverse of the frequency response of the monitor loudspeakers could be seen in the spectra of the recordings [released recording that is]”.
To recapitulate, in order to assess the final sound to be released on some medium, which will be listened to by consumers on their listening setup (DACs, amplifiers, loudspeaker and room) some initial setup is needed in the stages of the processing chain (DACs, amplifiers, loudspeaker and rooms). Even selecting, positioning and “equalizing “ microphones requires some monitor speakers. In a more general sense, since some equipment is needed to produce any sound to be consumed and/or assessed on some other equipment, a chicken-and-egg like loop is created: speakers are evaluated using recordings which are made using microphones which are evaluated using speakers. Toole calls this loop “The Circle of Confusion”. The consequence is: “… if consumers are to hear what the artists, musicians and recording engineers created, they should have similar loudspeakers and rooms. If not, they will be hearing something different”.
So, how can the loop be broken? While there will always be a need for some processing in the feed-forward path, a few things can be done to alleviate if not eliminate the effects of the feedback path in the loop:
- use science, psychoacoustics and the engineering principles of breaking up chains into links and assess each link with processes of higher sensitivity; the result should be well characterised components.
- make an effort to produce an output that is less correlated with the performance and frequency response of monitoring equipment; the result should be sound closer to the original ones.
- standardisation of the specification of equipment beyond simplified versions of +-3dB bandwidth of speakers.
If these measures are observed, there is perhaps a chance that in the future released recordings are less affected by the monitoring equipment used during production and to a controlled degree by the processing equipment on the path from microphones to medium, like in the idealised version.
Post Scriptum
As an engineer, the “circle of confusion” makes me immediately think about a very common issue in engineering: manufacture some product to a certain specification and precision, perhaps even a simple thing such as a ruler, requires tools that are themselves manufactured to a certain specification and precision using other tools, they in turn manufactured to a certain specification and precision using other tools, etc. traditional engineering has practices for that problem, e.g. using oversampling and statistics to characterise measurement tools.
