Digital recording in a home studio

Maximising Quality

When recording performances (music, spoken word, etc.) the aim is normally to capture the best recording possible from the start. This typically means using good quality microphones, pre-amps, and high-quality recording mediums.

In the analogue world the de-facto recording medium was tape, with professional studios using reel-to-reel tape systems costing many thousands of dollars, and would purchase good quality tapes. In the digital world, the medium is data, and the equivalent of the tape machine is the Analogue to Digital Converter (ADC). In a home studio environment the DAC is normally built into the sound card.

In order to maximise quality a good quality sound card should be used with decent pre-amps, low noise, and a reasonable quality ADC.

But to maximise quality, it is important to follow some good recording techniques.

Resolution and bits

Audibility

In the digital world, it is important to consider resolution. Normal full-resolution digital audio follows the "Redbook" standard which defines the format used on audio CDs: 16-bit resolution, 44.1kHz sample rate. This resolution was based around the principle that it fully captures all audio that can be heard by humans. That principle stands today. There are some who claim otherwise, but their claims defy the substantial weight Science and Engineering. They are extraordinary claims which, therefore, demand proof. To date no such proof has been given.

For further information on this subject, I recommend the Xiph article 24/192 Music Downloads ...and why they make no sense.

16 bits of resolution, when used properly, supports a dynamic range of up to 120dB, which will let you listen to music so loud that your ears bleed, whilst the quietest possible sounds are masked by the thermal noise in the components in your amplifier.

Why you should record at 24-bit resolution

When recording, it is recommended to record audio with a 24-bit resolution. If this additional resolution isn't audible, why is this?

To understand this, we need to look at the recording process.

One of the most difficult parts of recording audio is maximising the quality of the recording when recording to a medium with a limited dynamic range. When capturing audio we want to capture even the smallest nuances of sound and so we want to capture a wide dynamic range, but this is difficult when dealing with everything from loud drummers to soft vocal nuances. Traditional analogue tape machines had a relatively low dynamic range of around 60dB (top end systems could get to 80dB) and getting the recording level set to maximise the quality of the recording was always a challenge.

For digital, 16 bits is enough to capture the full range of audio and exceeds the dynamic range of tape. However, it is only just enough, which means that a slight error in setting the recording level can result in loss of quality.

This is compounded by the fact that digital ADCs are far less forgiving of errors than their analogue equivalents. If you set the recording level a bit too high with tape, the tape naturally compresses the audio and it still sounds good. By setting the levels so that VU meters peaked occasionally, you maximised the dynamic range without significantly compromising the audio quality. With digital, if your signal peaks, it has a devastating impact on the audio: it simply chops the top off the audio!!

Trying to maximise the recording level with digital recording is a very, very bad idea!!!

The generally accepted rule is to record so that your peaks are 10-12 db below the maximum level (known as 0dBFS). This prevents any unexpectedly loud peaks from hitting 0dBFS and getting chopped.

However, by doing this, you are reducing the amount of dynamic range you can record by 10-12 dB.

The answer is to increase the dynamic range of the recording system. We can do this by recording in 24-bits. This increases the dynamic range so much that you can actually turn the recording level down far below this, almost to the point where you can't hear the audio on the recording, use digital gain to make the signal audible again, and you shouldn't lose any audio quality.

Simply put, using 24-bit to record removes any issues of dynamic range in the digital domain, which makes making good quality recordings easier.

You still have to consider analogue

Although recording in 24-bits removes the dynamic range constraints from the digital part of the recording chain, prior to the ADC there are a lot of analogue components that still need to be considered, and optimised. As a minimum, there will be some sort of analogue audio pre-amp before the ADC input and if this isn't fed with a good signal it will result in poor quality recordings. There may also be microphones and mic type, directionality, placement, and room acoustics all play a part in this as they always have done. Just because using 24-bit to record solves one part of the recording equation, that doesn't mean you can ignore the other parts.