Bob Stuart, creator of MQA, talks in detail about this revolutionary British technology that sets a new standard in capturing, delivering and reproducing digital audio.
White Glove: Fairytales – (Appendices)
Appendix: Technical measurements
|Analogue–analogue frequency response of the MX-80 machine. Note that the response droops slowly by 1dB between 12 and 20 kHz. The roll-off above 20 kHz is gentle and minimum phase.|
|The corresponding impulse response.|
|In Red, the L channel noise floor; Green is a 16-bit dither level for comparison. We see that the system achieves a noise floor equivalent to 15 bits above 10 kHz, somewhat lower below.|
|A snapshot showing the improvement made by correcting ADC mis-alignment. This is a spectral analysis of the nominal 1 kHz test tone. In Red we see that the harmonics (the oscillator was not low distortion) and signal confounded by bit errors. Yellow shows the result after correction. In-harmonic converter distortion is reduced by around 40dB.|
|Illustration of the quetzal nature of room pitch can be seen in the wavering lower line in this interval between two notes (taken from the digital extraction).|
Appendix: Timeline of work
This section is a compilation of notes from Thomas, Morten and Bob.
- 1982 recorded at 50.4 kHz (turns out to be 50.35)
- Pre 1985 Mastered to Betamax (Sony F1) and digitally transferred to CD.
- 2005 Outakes from master tape transferred to 88.2 kHz by Morten at Sony (Battery Studios)
- 2010 Master tape is located, however the viable machine at Sony is lost
- 2015 April tape is played back at wrong speed/sample rate with original machine. Recorded from analog out.
- 2015 July Playback of original/analog master of My Funny Valentine shows the pilot tone to be 982 Hz not 1 kHz. Info is passed along (this measurement led to confusion)
- 2015 December release turns out to be from preview files, wrong pitch and a great deal of jitter/artifacts
- 2016 MQA requested to help.
- 2015 Bob shows that a pitch translation is viable for the digital extractions
- 2016 January test from digital out reveal the ADC and clock as another source of artifacts.
- 2016 February Info is received from LA, hopes of restoring the machine to original state, and completing a stable digital transfer.
- 2016 April Attempt at rebuilding fail. Contact with Battery reveal their machine as the same as used in 2005. Test tape is sent. Turns out the machine has broken down. Using the digital out cable, it turns out their backup machine is a 48k machine. Sending the tape and cable to various studios in US, it turns out all machines where in fact 48k machines (even though some thought they were 50.4 machines). Sony holds the test tape, but have postponed the work indefinitely as they have no complete schematics or detailed service manual.
- 2016 June Attempt using the National Libraries x-80 machine for a digital transfer fails due to excessive read errors.
- 2016 August work started at MQA using the various files
- 2016 November 2016 draft completed.
- 2016 December, listening session with Arild
- 2017 March. Final draft submitted to Label for review
- 2017 July, preparation of deliverables for vinyl, CD, download.
- Performers, Producer of the original recording and Label: Radka Toneff, Steve Dobrogosz, Arild Andersen;
- Thomas Bårdsen (Nasjonalbiblioteket, Mo i Rana), master transfers.
- Bob Stuart (MQA) , project oversight, audio recovery and restoration
- Peter Craven (Algol/MQA), measurement analysis and design of software tools
- Morten Lindberg (2L), 2005 transfers and Remastering this edition on behalf of MQA.
- For their many efforts on the recovery at Norway National Library in Mo i Rana: Thomas Bårdsen, Geir Iversen, Svein Vatshaug and Rune Sund, Nordmark;
- Andreas Risanger Meland (Grappa);
- Friends: Morgan Nicolaysen, Pål Bråtelund.
Appendix: Dynamic Range
Analysis of the dynmaic range of the earlier and MQA releases.
|2017 MQA Master||2017 MQA vinyl master||MQA 24b
|MQA CD (decoded)|