It is hard to believe that there are sound production software programs on the market that use MP3 files as their file format for editing and recording audio. What follows are the reasons why MP3 should never be used to record, edit or mix with, what the MP3 codec is and what file formats you should use for recording.

The MP3 codec (compression - decompression) is a lossy audio format. As you know, mobile phones, ipods and webpages all use MP3 for audio playback and distribution. The problem is that MP3 loses audio data and so is not very useful for producing music of high quality.

The reason it has so widespread use is that, as a lossy format, it can achieve compression ratios of 10:1 and so is useful for file distribution across the web (web pages with MP3 audio open quicker) and storage due to the decrease in file size that it achieves (you can save more songs on your hard drive or personal music player.)

However, a negative aspect is that, due to the compression process, audio quality is affected.

What should you use?

Recording, editing and mixing

Use WAV (Microsoft Wave) files or BWF (Broadcast Wave) as they are cross-platform compatible and industry standard. These should be at least 16 bit, 44.1 kHz red book CD standard audio files.

Storage

Use WAV or BWF. If you want to send a lossless file to someone, you can compress the file using FLAC which results in a 50% - 60% reduction in file size. FLAC, however, is not supported on most media players. FLAC also reduces audio quality. FLAC is used quite frequently for digital file transfers of hi-fidelity recordings such as classical pieces. There are free FLAC encoders/decoders available on the net.

Distribution

For giving CDs to people, you should always burn your uncompressed master WAV or BWF files. Never MP3 as you always want to provide people with the best quality available. MP3 is okay for web distribution if you have to compress your audio, but try to use the highest bit rate possible. The best data transfer rate is 320 kbps which can be perceived as being CD quality on many systems but 160 kbps is definitely the lowest you should go. If your sound production software of choice does not allow you to save as an MP3, there are free programs such as Audacity and LameXP which can compress your masters to MP3 if you wish.

How does it work?

The MP3 codec works using psychoacoustic (study of interrelation between the ear, the mind and sound wave) phenomenon called simultaneous (auditory) masking and temporal masking.

CD audio stores more data than the brain can process such as two notes that are close together being perceived as being one note. If two notes are similar but one is quieter than the other we may not be able to hear the quieter note.

MP3 encoders analyse the incoming sound signal, breaks it up into mathematical patterns and compares these patterns to psychoacoustic models stored in the encoder. The encoder then gets rid of most of the data that doesn’t conform to the psychoacoustic models. When you compress a WAV to MP3, for example, you can determine the level of data that is going to be discarded by entering the data transfer rate in kbps. The lower the kbps data rate, the more of the original data will be discarded and the greater the loss in audio quality.

Simultaneous masking

Simultaneous masking is when you can hear the sound of a guitarists fingers on the string during soft passages but can’t hear the finger movements when the guitar is played loudly. This is due to the relationship between frequencies and their relative loudness levels.

To explain this in MP3 terms, you have a 1 kHz sine wave and add a second sine wave at 1.1 kHz, 10 dB quieter. A lot of people will not be able to hear the second sine wave. This is due, not to the difference in loudness but because the two tones are close to each other in the frequency spectrum. When you increase the frequency of the second sine wave, keeping the loudness the same, you can hear the two separate tones clearly.

What the MP3 codec algorithm does is give less disk space to the audio that we cannot hear and give more disk space to the dominant signal.

Temporal masking

As well as simultaneous masking, there is also temporal masking. This is based on time. Humans have problems hearing sounds that are close together. If a loud sound and quiet sound are played at the same time, we can’t hear the difference. If they are played at an interval of say 10 milliseconds, we are able to differentiate between the two sounds. The MP3 algorithm quantifies whether the time differences make a difference and, therefore, decides whether to throw the data away or not. It also works in reverse if a quiet tone plays before a louder tone so both premasking and postmasking are accounted for in the algorithm.

Bit rates

When encoding an MP3 file, the user can decide on the amount of bits to be given to data storage. The unneeded parts of the signal are decided by the mathematical algorithm of the MP3 codec (masking requirements) and the bitrate. The bit rate is the number of bits per second that is given to storing the final MP3 file. The higher the bit rate, the better the audio quality, but the storage needed on your hard disk is greater. Small files mean less audio quality.

You can use a constant bit rate when encoding or a variable bit rate. Variable is better as music is built up of different dynamics so if you have a constant bit rate of 320 kbps then all loud and quiet parts of the music are going to be treated the same. Variable encoders and decoders vary how much bits they are allocating depending on the quiet and loud parts of the audio you are encoding or decoding thus resulting in smaller files than if a constant bit rate where used. There are, however, disadvantages with using variable bit rates. They might not be able to be played by old MP3 players and there may also be timing issues when playing back your audio.

Bit rates and sample rates

In professional music recording software, Bit rates are a measure of the amount of data stored for every second of audio, whilst sample rates determine the resolution of the audio by how many audio samples are taken each second (measured in kilohertz, kHz.) The more samples that are taken every second, the more detail there is in the audio but the more storage space is used.

There are pros and cons regarding the MP3 codec as outlined in these two articles. The main aspects to take away are: Use computer music recording software that implements WAV files for recording and use MP3s for casual listening.

You do not want to be using, for example, a beat creator that uses MP3's as your music will just sound thin. If you are looking for a program to make beats with, then I suggest you look into DubTurbo as it uses .wav files.