These technical recommendations concerning your preparatory work for the recording, mixing, encoding and transferring of each source file to our mastering studio...
A "good mastering" is achieved with a "good mixing". And a "good mixing" is achieved with a "good recording"... The intrinsic properties of the source largely determine the final sound quality!
If you don't go to a studio, record and mix yourself every instrument at the "good level" and the "good position" in the stereo. Your music has to sound already correctly (balanced mix) without treatments on the master bus (main stereo) of your mixing desk.
Add, as appropriate, treatments and effects on the channels: equalization, compression, distortion, chorus, phaser, flanger, reverb, echo, etc... Avoid stereo effects on instruments with low frequencies (< 90 Hz). Consider using sub-groups (busses) of your console. If there are recoveries of frequencies in your mix, adjust the instruments which interact in different frequency bands.
By contrast, no filtering and no limiting on the master bus. Keep amplitude in dynamics because excessive compression limit our possibilities of intervention and increase the sensation of crushing sound. If you put a compressor on the master bus, do not exceed 2 dB of gain reduction and preserve the transients. Do not try to "boost" your mix, let our sound engineer choose the tools and parameter settings.
Check the signal level through the peak meter on your mixer (level not too low or too high). In digital, the maximum level should be close to 0 dB FS without overload. It's recommended to leave a small margin (peaks between -1 and -3 dB FS). Check the compatibility of your sound in mono listening by the mix of the 2 channels. As the volume will change during mastering, it may be preferable to place the fades at the end of the audio treatment (especially long fades).
Prefer listening on 2-way speakers at minimum, avoid mixing with headphones. Do not make too long mixing sessions because your ears quickly get used to defects. Take regular breaks, and then listen again your work!
During the mixing, it may be a good idea to clean up your recordings with low-cut filters, but be careful with instruments that need to keep bass frequencies (like the bass line, or the bass drum). Once it's cut, it's cut!
If you have done all your mixing with processings on the master bus (equalization, compression, limiting...), they are part of the mix! Since you adjusted your levels by listening to them, you can not remove them like that. It is therefore preferable to resume your mixing, and edit or delete these processings. This operation will have a greater or lesser impact on the balance of sounds and transients. Remove at least the limiter (at the end of the chain) and check carefully the result...
In your computer software, export the audio mix ("bounce") to the project sample rate (eg 44.1 kHz) and the internal resolution of program (eg 32 bit float).
Observe the following points:
The sampling frequency of 44.1 kHz is usually used in music (as for the Audio CD). The sampling frequency of 48 kHz is usually intended for the audiovisual world (video and film). But currently, with HD or Studio quality files, we recommend you working from the beginning of your project with a double or quadruple sampling frequency. The best results are obtained with a complete processing at a high sampling rate (recording, mixing and mastering). The sub-sampling at the end of the processing is of excellent quality.
As most software uses an internal quantization of 32 bit float (see 64 bit float at present), the result of your mix is a priori in 32 bit float if you have not applied dithering. 32 bit float is a very good resolution for sending your mix. It allows digital processing after export without significant reduction of quantization. The 24 bit is also interesting for sending your mix because the dynamic range remains important. This resolution is that of most digital / analog converters. If your software propose you to apply a dithering when exporting to 24 bit, it is because its internal quantization is a priori greater. If you decrease the resolution during export (to 24 bit or 16 bit), you must apply a dithering (UV22, POW-R...).
The rules for the mixing and the encoding of the files are the same as above, except that you will export your mix in multitrack, ie in several stem tracks (or sub-groups, or busses), so that the sound can be tweaked in depth. Stems are a set of pre-mixed groups of tracks from a song...
Export from your computer software 8 stereo stems (maximum) with treatments and effects: 1 file per stem, and identical timecode between files (same duration). Draw together the instruments according to their transient (percussion, synthesizers, voice...) and the interest as there are to separate some sounds for a new mix. If the recordings do not have the same starting point, place a short synchronization sound at the beginning of each stem (eg snare drum). Then, verify that the sum of these stems reproduce your original stereo mix that you have to send at the same time for control.
For example, organize your files as follows:
Export your files without applying dithering because the background noise of each stem will add up and will be too great after mixing. With the number of stems, the background noise increases. So choose a resolution of 32 bit float, or 24 bit without dithering.