In order for your mastered tracks to have that sparkle and clarity that you envision, it is important to follow some mixing guidelines. For these discussions we'll assume the project is for CD-Audio, but all the basics apply no matter what the final delivery medium.
First we'll address digital audio file formats, sample rates and bit depths.
1.) DO NOT SAMPLE RATE CONVERT. If you are mixing 'in the box' or through an external digital mixer, the mix files should be at the same sampling rate as the session. That means NO SAMPLE RATE CONVERSION. Hopefully you tracked at 48k or ideally 88.2k or 96k, but many engineers make what seems a logical assumption that since CD-Audio has to be 44.1k, then they should convert their mixes to 44.1k even though the session was tracked at a higher sample rate. There are three reasons not to do this. First, for ultimate sound quality the project should be mastered at the higher sample rate at 24 bits and then downconverted to 44.1/16 as the final step before the Production Master CD or DDPi is cut. Second, it is likely that a real mastering room has a much better sample rate converter than those found in even large studios. And third, there is a growing trend of artists and sites offering hi-resolution downloads, so it is advisable to have the mastering room create a separate data master containing the mastered 24 bit files at the higher sample rate.
2). MIX AT 24 BIT RESOLUTION. Even though your session may have been tracked at 16 bit, it is important to mix at 24 bit resolution. The reason is that if you do anything more than play one sound file at unity gain out of one buss, even something as simple as changing the gain by 1/10th of a dB, your system has to expand to a 24 bit word length internally. So if you output your mix as a 16 bit file, you will create truncation distortion and lose fidelity. Since hard drives are so inexpensive these days, you really should be tracking as well as mixing at 24 bit resolution.
In the mastering world we are seeing a disturbing trend of people wanting to deliver mp3's as sources. mp3 is barely acceptable as a consumer audio format, and is certainly not suitable for mastering.
CAVEAT: Know Your System! If your DAW or digital mixer operates at 32 bit or 32 bit floating, you need to apply 24 bit dither if you are outputting 24 bit mix files. Or you should output them as 32 bit files and inform the mastering room of this.
Relating to points 1 & 2, There is no 'getting back' quality. If you have created 16 bit mixes, loading them into a program that will output them as 24 bit files will do nothing for you. All it will do is add 0's to the lower eight bits. The sound quality will be the same but the file will just take up more disk space. Likewise upsampling a 44.1k file to a higher sample rate will often decrease quality except in the best of sample rate converters, and then the quality will likely be the same, it will again just take up more disk space. If all you have are mp3's of your mixes, converting them to .aif or .wav will not improve their quality. In fact, you will lose quality with such 'transcoding'. If all you have are mp3's any qualified mastering engineer will know how to handle them to result in the best quality possible. The same is true if all you have is an audio CD-R for a master - it's not doom, many great records have been made from 44.1/16 masters.
3) AUDIO FILE FORMATS. All mastering rooms should accept any of the standard file formats - .aif, .wav, .bwf or .sd2. Normally interleaved stereo files are preferred at the mastering room, but split L/R is fine and there is no difference in fidelity. Also, keep your file names under 15 characters. It's not necessary to name the file with the full song title. Just the first two or three words of the title is fine. See the documentation topic below.
4.) DELIVERY MEDIUM. Most mastering rooms should offer you the option of uploading your mixes. You should put all of your mix files into one folder and then zip that folder prior to uploading. Zipping a folder will perform and internal checksum which ensures that nothing was corrupted during the upload/download. If you want to deliver your mixes on disc, make sure you are creating either a data CD-R or DVD, not an audio CD-R. If you have tracked and mixed at 48k or above at 24 bits, and then you put your files onto an audio CD-R, you have lost fidelity. The reason is that an audio CD-R, like CD-Audio, can only be 44.1k/16 bit, so even though you may not realize it your CD burning software has converted your files to 44.1k and then dropped the lower 8 bits, and you have lost fidelity through both an inferior sample rate conversion and the file has been truncated. On a data CD-R or DVD, the audio files can be of any sample rate and any bit depth. Plus the method in which a data disc is written is much more accurate than an audio CD-R, so there are no errors on the data disc. Your 24 bit higher-sample rate files are much larger than CD-Audio files, so an album's worth will fit on one DVD-R, but will require 2 or more CD-R's, which is fine.
Any mastering room should also accept your sources on a Firewire or USB drive.
5.) MULTIPLE SAMPLE RATES. Do not mix sample rates unless clearly noted. Mixed sample rates often occur when multiple studios and/or engineers have been involved in a project. One engineer may have tracked and mixed at 48k, another at 88.2k etc. If you have a situation where your mixes are at different sample rates, put them into separate folders indicating the sample rate of the files they contain. Bit depth does not matter, if some of your mixes are 16 bit and some 24 bit but all are at the same sample rate this will not cause a problem at the mastering stage.
6.) EDITING AND OTHER PROCESSING. It is important that you deliver the first generation mix files for mastering when possible. They should not be loaded into any program to convert .aif to .wav, or split L/R to interleaved or any other seemingly transparent process. They ESPECIALLY should not be loaded into any workstation for normalizing, fades, edits etc. Your mastering engineer can easily perform fades and edits, and normalizing should not be done in any case. In most cases eq, compression or limiting should not be applied across the stereo buss while mixing. We'll address that in the next post.
CAVEAT: If your project will require a lot of editing, sometimes budget will not allow editing in the mastering room. If you do your editing in the studio, just be sure that everything stays 24 bit, the sample rate stays the same and that no level changes are done if at all possible. If your editing requires a section of a song to be raised or lowered in volume, then 24 bit dither should be applied when the finished edit is output.
7.) DOCUMENT, DOCUMENT, DOCUMENT. Too much information is a good thing. Your sources should be delivered with documentation indicating the sample rate and bit depth of the files as well as the title of the file and the corresponding full title of the song. It is also helpful if your mix engineer can provide details of whether the mix was done in the digital domain or through an analog mixer, if analog tape was used at any stage and whether stereo buss processing was applied and if so what kind and how much was applied and if any dither has been applied at any stage.
8.) ONLINE DELIVERY. If you are going to make your finished project available through download only, do not assume that you can lower the quality of any steps in your production. Please read our first blog entry Audio Production in the Age of mp3 and AAC for details.