Signal to Noise Ratio (SNR) – How it can destroy your recording

What you see in audio theory does affect you – this is proof. How should you reduce background noise to record the signal better? That’s what we’ll be explaining in this article. Signal to noise ratio is the proportion between signal strength and the noise floor effect, the so-called background noise. This problem can arise because of interference in the electric current, heat, vibrations or even poor-quality equipment. In-the-box recordings aren’t an exception. The quality of the analog-to-digital conversion is fundamental to avoid quantization errors, which cause noise in the digital file. SNR also applies to acoustic, electric, optical and digital signals. It’s adopted in other telecommunication fields too, such as engineering and science.

The signal to noise ratio is the proportion between the dynamic range and noise floor.

Signal to noise ratio

SNR (signal-to-noise ratio) is the comparison between the highest signal level emitted and the noise that comes out from the outlet of the equipment. This term represents the proportion between the strength of the unwanted noise and the audio signal emitted. The greater the distance between them, the better the signal to noise ratio is. Thus, the signal will be sent with quality, no distortions and the noise won’t be perceived. Or at least it won't be strong enough to interfere in the sound. Most professional audio equipment will tell you what the Dynamic Range levels and SNR (signal to noise ratio) are. Dynamic Range is basically the variation in dB the equipment is capable of receiving. Example: 100dB of Dynamic Range. In digital equipment, that means it can withstand signals from -100dB to 0dB, since that’s the highest number in the dBFS scale, the one used in the digital field. We calculate SNR by subtracting the noise floor from the value of Dynamic Range. Example: 70dB SNR. For a piece of equipment with 100 dB of Dynamic Range, there must be about -70dB of noise floor.

Signal to noise ratio calculation

The signal to noise ratio is measured in decibels. We’re talking about a logarithmic scale, so this is how you do the calculation:

What is noise floor?

Background noise, also known as “noise floor”, can be caused by many factors, whether it be components from the equipment itself, impedance, electric current, signal interference, reflection in the ambiance, wind, vibrations, variations in temperature, humidity, etc. That’s why there isn’t one ideal formula to fix the problem. The best thing to do is to surround yourself with quality equipment in a prepared environment and always make sure to pay proper attention to this phenomenon in each production. Every audio-related equipment produces noise. Microphones, amplifiers, speakers, headphones, receivers, CD and DVD players, radios, record players, interfaces, sound cards, tables, telephones, smartphones, musical instruments, and even cables! Some are insignificant, others more perceptible. These noises get stronger with deterioration, age or mishandling of the equipment. Besides, as mentioned before, other things can interfere, such as noise from external sources. When playing and especially when recording, they make a lot of difference. A noise common to every equipment is the humming sound: constant intensity, few frequency variations. It's the famous humming that sounds a lot like a fridge.

Want to understand this better?

• Turn on the guitar amplifier without plugging the instrument;
• Raise the volume.

Did you notice the sustained “hummm”?

Next step

• Now, plug your guitar to the amplifier. 
• Raise the volume of the instrument to the max and play something – preferably a chord and let it resound. 

The resulting sound will contain the chord you played and the humming sound on the background. If there's almost no humming, great; that means the SNR is high and there won’t be noise interference. The difference between the signal and the noise floor is small, even if the guitar is at its maximum volume. But if the noise is still there, strongly present in the sound that comes out from the amplifier, the SNR is low. The difference between the signal and the “noise floor” is small, even if the guitar is at its maximum volume. The most obvious and immediate thing to do is to raise the volume of the amplifier, right? No! Because when you do that, you raise the sound of the guitar and the noise together. SNR is measured in decibels, a measurement defined by logarithmic calculations. That’s why only raising the gains in the sound device isn’t the perfect solution. When you raise the overall gain, the volume of the noise is raised proportionally to the signal.

Understanding the signal to noise ratio in digital recordings

Now let’s talk about how this relates to the digital world. In recordings, not having control over the signal/noise ratio can ruin your work. That small, even “tolerable” noise makes all the difference in live playbacks and performances if it isn’t eliminated in the recording process. Nowadays, almost all digital sound devices used in professional recordings are made to mute their sound outlets when there’s no signal coming in. In this situation, we see a very low SNR. Problem solved, right? Wrong! To transform an analog sound into digital data, its interface captures thousands of moments of sound in a second. In a CD, every second of sound corresponds to 44 100 captures or 44.1 kHz. This is known as the sample rate. It’s as if the digital converter took pictures every second the sound wave comes in. The more “photos” in a second of sound, the better the sample. The samples are stored in bits. The bigger the bit depth, the more faithful it is – the resolution and quality are also better. For instance, the CD has a value of 16 bits per sample; the DVD and the Blu-ray have a value of 24 bits-per-sample.

Quantization errors and the signal to noise ratio

But sound waves vary all the time. The converter may interpret these variation incorrectly. Digital reading parameters don’t always perfectly fit the analog waves sent to it. When a wave doesn’t fit with a parameter in the digital scale, the converter tends to round up the analog value to the closest corresponding level in the digital world. This deviation is known as quantization error. It’s one of the factors that create background noise in the digital system of the recording. There is a conflict between the nature of the analog signal and the digital data representation. So, even if the sample rate is optimal and the bit depth is big, this error can happen during conversion.

How do you balance SNR in the analog/digital conversion?

By improving the quality of the recording. Every bit added in a sample increases 6 dB in the dynamic range. For instance, the dynamic range value in a 16-bit recording is 96dB. By applying 24 bits per sample, you get 144dB on the dynamic scale. The greater the dynamic range, the more space you have to work. When working with 16 bits, you have the number 2 elevated to the 16th power of possible levels of quantization (65 536). With 24 bits, you have the number 2 elevated to the 24th power (16 777 216). The more “steps” you take to the represent a digital sound, the less roundings there are during quantization Most converters nowadays work with 24 bits of depth. 24-bit files not only occupy a lot of space in the HD, but they also require a lot from the DAW processor and RAM on your computer. It’s already possible to record in 32 bits. This option introduces the floating-point, which offers 8 more bits of volume. This is ideal to increase headroom and to be able to use the compressor, but it still takes a lot of space in your hard drive. However, 24 bits is already the standard, so it isn’t advisable (or well looked upon) to record in fewer bits than that.

Dithering

Dither exists to mitigate digital noise created by errors in quantization. It adds a very low background noise on the first bit, the lowest quantization level the system recognizes. This phenomenon masks the noise created by quantization errors. Therefore, with 24 bits, the dither is at -144dBFS; with 16bits, it’ll be at -96dB.

Beware of THD!

THD – Total Harmonic Distortion – is another important parameter to measure the quality of the equipment, along with the SNR and Dynamic Range. It determines what the possible level of distortion in the equipment is by comparing the audio signals that go into and come out from the device. Distortion happens because the electronic components of the equipment aren’t linear. As such, it produces small harmonic deviations in the output, especially when it receives a high-intensity signal. The lower the THD rate is, the less distorted the sound. Most mixers and pre-amps present a THD rate of 0.01% or lower. Amplifiers usually bring the THD up to 0.5%. The THD may vary according to the frequency – the higher the frequency, the higher it’ll be. To sum up, when recording with high gains, pay attention to the THD!

Tips to improve SNR in your recording

• Don’t record very low signals. It isn’t necessary to record that close to 0dB since you'll be working with 24bits. Make sure, however, that you are working within an appropriate range, with peaks between -20 and -10 dBFS;
• Set up the gains of the whole chain to operate in the best working space (from the instrument to the DAW meters); 
• Avoid long cables. The distance to the input of the device may cause loss of signal and noise;
• Use the direct box to capture directly from the line;
• Be on the lookout for external interferences – vibrations coming from the floor, equipment sitting too close to each other, continuous use, how hot the devices are, electric circuit (some LED lamps interfere with signal); 
• Make sure to use the best corner of the room to record; 
• Take a look at the bit depth and the sample rate before starting the project;
• Always leave plenty of space in your computer’s HD;
• Apply dithering if necessary. 

And of course: always try to use good equipment. As you may realize by now, signal to noise ratio is something to be aware of. There is no perfect formula, only careful signal calibration. It’s a result of the sum of compatibility, quality, and cohesion in the structure and components involved in the recording. It requires observation and control over the whole process. Now go and have a good time recording!