Tony Miro is a Mastering Engineer out of the U.K. I've had several discussions with him on mastering and he wrote up this article in response to some of my questions about it. Thought I would share it.
The original article can be found here:
http://www.guitarmasterclass.net/guitar ... opic=35566
------------------------------------------------------------
Thought that it might be helpful to discuss this now as there's been some interest. If there is enough interest and people want it I'll speak to Kris about possibly setting up a separate sub-forum for Mastering...
[I should add that the following is what I do and what most professional MEs do as well. I'm not claiming though that everyone who says they are an ME does the following...]
Mastering and what a Mastering Engineer (ME) normally does
Mastering is the final phase of audio production prior to the production of a glass master replication disc or the digital distribution of a mastered digital audio file. Mastering involves several distinct but over-lapping processes: critical listening, sequencing, processing, quality assurance/checking, production of master CDA/DDPi/Digital file. These processes may take place once, more than once or may not be required depending on the specific audio project and client's wishes and requirements.
Critical listening
Critical listening takes place one or more times during mastering.
Firstly, before any mastering can take place the audio project must be listened in an objective and critical manner. The ME here determines a number of things:
Is the audio ready for mastering or does it need to be re-tracked and/or re-mixed. The ME may comment on the quality of playing on the audio if they feel it undermines the overall project - for instance, if an instrument is out of tune and/or there are timing issues. The ME may also comment if they feel that the recording is not suitable for the respective genre. The ME will almost certainly comment if there are tracking and/or mixing issues especially if these involve unwanted distortion, incorrect gain staging and inappropriate use of processors on the 2 bus, audible artifacts, stereo field, mono, phase and DC offset issues;
If the audio is suitable for mastering what sequencing and processing may be necessary so that the mastered audio achieves it's full potential and/or meets the client production criteria. Production criteria includes what the producer wants from the project - for instance should the audio emulate a particular recording and/or do some elements need additional emphasis - and what the end-use will be - format (CD, DDPi, SACD, MP3, etc), type of playback (consumer audio, radio transmission, film/theatre, etc)
It is only after extensive critical listening and discussion with a client that an ME will decide what processing, if any, may be required and what hardware and/or software to use and in what order. There is no 'magic bullet', no 'must do/have' processing, no 'pre-set' in mastering as everything depends on the individual audio and the production criteria.
Secondly, since mastering affects an entire stereo track any change made by an ME to one element may affect other elements of the stereo file i.e. changing the sound of a particular instrument/vocal should not be to the detriment of the entire track. So critical listening will take place at discrete points during mastering and will involve a number of A/B comparisons between the original audio and the contemporaneous mastered version to ensure that any changes benefit the audio overall.
Finally, critical listening will take place at the end of the mastering session and involves listening to the master CDA/DDPi etc that has been produced/burnt by the ME. The ME here does two things: first, the master copy is listened to in its entirety to ensure that there are no audible errors or issues with the CDA, etc; Second, the CDA will be played on a number of different cds to ensure both that translation of the audio is as wide as possible. Translation ensure that the cd can be played, and sounds appropriate on as wide a range of different CD players as reasonably possible ranging from ghetto blasters and car stereo systems through to high-end audiophile set ups.
An ME must be able to hear audio from 20Hz to 22000Hz accurately so critical listening requires the use of high quality analogue/digital and digital/analogue converters, mastering grade mid or far field monitors and good room acoustics. Put simply, if the ME can not hear the audio accurately they cannot make an informed decision on what, if anything, to do.
Sequencing
Sequencing involves both ensuring that the tracks that make up an audio project are placed in the correct order and flow together properly to sound like a coherent and cohesive work. Individual tracks are brought together in an order determined by the ME and/or producer to achieve appropriate balance and impact. Individual tracks may have specific processing, internal fades and automation applied to them and in addition cross fades may be applied between tracks. Sequencing between tracks may involve specific adjustments to track levels to ensure that all the tracks form a cohesive project and that tracks there are not unintended shifts in volume from one track to the next. Generally level changes made at this stage do not equate with, and are not achieved by, normalisation. The adjustment of track volumes is often done by careful A/B comparison of the perceived volume of the main vocals in each track. In the case of Red Book CDA masters tracks will be sequenced so as to include silent gaps both at the start of the CDA and between tracks and any hidden tracks and count down/up flag timers will be added as required. An accurate PQ file will be produced that contains track order, numbers and duration. Once the entire project is sequenced any overall processing may be applied before the master CDA/DDPi etc is produced. This may include, for instance, a particular EQ/Compression mix to achieve a uniform vibe and any dithering prior to burning the CDA.
Processing
Contrary to popular belief effects and dynamics processing may not be the main role of an ME and, in the case of audio that has been particularly well mixed may not take place at all. Where processing does occur, and as already mentioned, it generally is done to improve the overall sound of a track rather than a single element within a track. This means that processing normally is done on the 2 bus, supplemented by additional processing using mid/side and or L/LR/L- matrices. It often requires a high degree of precision and care to ensure that the processing does result in a deterioration of the overall audio quality; maintains the stereo field, level balance and phase coherence, and; does not introduce any unintended and unnecessary distortion or artifacts. Processing at the mastering stage normally requires the use of specific, mastering grade hardware and/or software. The processing most commonly used in mastering are EQ and compression followed by limiting. Contrary to popular belief multi-band (or split-band) compression and reverb are rarely used and then only to correct very specific issues with a particular mix.
Quality Assurance/Checking
As already mentioned an ME will listen to their work critically at a number of stages to ensure that the master improves on the original, has not introduced any unintended issues or problems and provides as wide a range of translation as possible. In addition to these quality assurance checks(QA) an ME will also ensure that their work meets any appropriate standard required by the client. In the case of a CDA master, for instance, the CDA will be produced to the Philips/Sony Red Book standard for an audio Compact Disc; for a Super Analogue CD (SACD) that it is produced to Scarlet Book, and so on. In addition to the sequencing standards QA here may also include the embedding of metadata in to the audio data. This metadata may include ISRC (International Standard Recording Code), bar coding, track titles and artist names and so on. QA will also include checking that the final CDA master has minimal C1, C2 and CU errors so that the CDA master meets the requirements of a professional replication plant's QA for the preparation of the glass master for print.
Production of the master Red Book CDA/DDPi etc
When an audio project has been fully sequenced it may then be: burnt to produce a master Red Book CDA, or; encoded as a digital DDP image, or; rendered as a high quality mp3 for itunes and so on. In the case of a Red Book CDA the ME will use a CD writer capable of burning ISRC and metadata in DAO (Disc at Once) mode AND probing the CD to produce a QA Check for C1,C2 and CU errors. DAO is used rather than TAO as the latter is not an acceptable burn format for glass master replication due to issues with PQ logging. DDP (Disc description protocol) involves the production of a digital image of the audio along with any additional files that describe the project's structure, table of content (TOC), index, subcodes, etc. This image may then be sent direct to the replication plant without the need for the ME to burn a master CDA. NB not all replication plants accept DPP images. Also note that very few DAWs are capable of DDPi production. Finally in the case of an mp3 for digital distribution via, for instance, i-tunes, an ME may embed any appropriate metadata and images in to the mp3 including ISRC. Note that mp3 encoders generally do not embed ISRC and so this must be done as a specific meta tag within the ID3 v2.4 sub-code.
