Boost Your Audio Quality with these Simple Recording Hacks
There are plenty of ways to spend money on improving your sound—an investment in better mics, better acoustics, better speakers, and the like can make a huge difference in overall sound quality. But there are also plenty of simple, inexpensive fixes that can improve the quality of your recordings.
Use the Highest Internal Resolution Your Host Program Allows
In addition to the recording resolution, most programs have a separate resolution for their internal mathematical processes that’s higher than the recording resolution (Fig. 1). This is because with digital recording, everything is numbers—and rounding off can affect the sound, so you want calculations that are as precise as possible. (As an analogy, you need only single digits to express 4 times 4—but you need two digits to express the result of multiplying these single-digit numbers.) With complex mixes, material with lots of dynamics like acoustic music, and long reverb tails, higher resolution can help lead to a more “transparent” sound.
Roll Off Low and Subsonic Frequencies
Some instruments generate frequencies below the lowest audible notes; for example, hitting a guitar’s pickups or body can generate “thumps” in the bass range. Mic handling noise, AC hum, and room rumble are also below the range of most instruments. Using a steep, low-cut filter (Fig. 2) to eliminate these low-frequency artifacts can “open up” the sound of a mix, but be careful: if the low-frequency sounds are inherent in the instrumental part (e.g., percussive acoustic guitar effects with flamenco), don’t get too aggressive with the filtering.
Use Quality Converters with Inexpensive Gear
Although many electronic instruments and effects include a digital input and/or output (e.g., AES/EBU or S/PDIF), these aren’t just for interfacing with other digital gear. A $500 signal processor isn’t going to have a $300 D/A converter sub-system onboard, but you can use the digital output to bypass the onboard conversion, and feed a quality digital-to-analog converter (Fig. 3). It really can make a difference. For example, when Korg did a virtual plug-in version of their M1 synthesizer, even though the plug-in used the same algorithms as the hardware, the sound quality was much better. This is because the original hardware used relatively inexpensive converters, but today’s audio interfaces use much better converters—and that’s what reproduced the sound of the plug-in.
Enable High-Resolution Modes on Plug-Ins and Soft Synths
To accommodate older computers as well as newer, high-speed models, many plug-ins offer a lower-quality mode that requires less CPU power (often called “Eco” mode), and a higher-quality mode that’s more taxing on CPUs (Fig. 4). Assuming your project can handle the extra stress, make sure you enable any available high-quality options, like oversampling.
Even if you have a super-fast computer, there’s another reason why low- and high-quality modes are useful. When tracking, choose lower-quality modes to allow for the lowest possible latency. When it’s time to mix, and latency is less of an issue, switch over to high-quality modes.
However, it may not be a good idea to use more latency than is necessary. Some engineers believe excessive latency causes timing issues that can smear the sound, and narrow the soundstage. Although I know of no definitive studies that back this up, there’s no logical reason to use more latency than necessary, in which case it doesn’t really matter if too much latency is a problem or not.
Use Noise Reduction, Even If It Doesn’t Seem Like There’s Much Noise
When you listen to individual tracks, you may not hear much hiss—if you hear any at all. But when you combine a lot of low-level noise on multiple tracks in a mix, the hiss can become audible.
Modern noise reduction algorithms, as used in iZotope RX7 (Fig. 5) and Magix Sound Forge, can reduce noise while maintaining sonic transparency. Fortunately, noise reduction is most effective on signals that don’t have a lot of noise anyway. So if the noise level is around, for example, -68 dB, then you can bring it down to -75 or -80 dB without any sonic degradation.
Spectral-based noise reduction requires a “noiseprint” of only the noise. You can usually find a suitable sample at a track’s beginning or end, or during “silent” sections. The program subtracts this noise from signal, leaving only the audio you want to hear.
This is the kind of improvement that seems minimal on a per-track basis, but can be significant in a final mix—it’s sort of like removing the dust from a painting. The painting hasn’t changed; it just looks a bit better.