Mixing and recording production dialog is a balance of capturing clear, direct speech with just enough indirect, environmental background sound for context. When there is too much background sound, speech can become difficult to understand. The balance between background noise and speech is referred to as the speech-to-noise ratio. Increasing the speech-to-noise ratio is beneficial not only for recording, but also for on-set communication channels or IFB feeds.
There are several ways to increase the speech-to-noise ratio, each with their own benefits and compromises. Let’s explore a few common techniques below.
1. Reduce Subject-to-Microphone Distance, Increase Microphone to Noise Distance
The single most effective way to improve the speech-to-noise ratio is by reducing the subject-to-microphone distance. Think about a live performance where a singer is holding the microphone directly at/on their mouth. That incredibly short subject-to-microphone distance can overcome loud stage volumes.
For this same reason, lavalier microphones are often used to capture production dialog in high-noise environments. Lavaliers can often be placed closer to the mouth than boom-mounted microphones, improving the speech-to-noise ratio.
2. Eliminate Background Noise Sources
When possible, it is best to eliminate acoustical noise sources before recording. For example, if there is a nearby air-conditioning unit, turning it off is far easier than trying to remove the sound in post-production. Similarly, if traffic noise is audible, moving to a new location is the ideal solution. Most of the time this is easier said than done.
Minimizing wind noise is also extremely important. Wind noise can wreak havoc on microphone preamplifiers and produce artifacts that may be impossible to remove. Be sure to use effective windscreens on microphones, especially when outdoors.
A quick listen with headphones often reveals acoustical sound sources, like wind noise, that are not otherwise obvious.
3. Use Directional Microphones
Microphones with cardioid, supercardioid, and shotgun polar patterns are less sensitive to off-axis audio signals, or signals that are not within the microphone’s ideal pick-up area. This trait can be used to reduce background noise. When noise sources cannot be eliminated, position microphones so that the noise sources sit within the null, or least sensitive part, of the microphone’s polar pattern. The exact null angle depends on the polar pattern. While a cardioid microphone null is at 180 degrees off-axis, as microphones get more directional, the null angle shifts. The null of a supercardioid microphone is at 127 and 233 degrees.
4. Use a Low-Cut Filter at the Microphone or First Stage of Amplification
A majority of background noise is ‘pink-noise-like’, with most energy at low frequencies and less energy at higher frequencies. Since there is very little voice energy below 100 Hz, using a high-pass (also referred to as low-cut) filter is good practice when recording dialog. For applications where speech intelligibility is more important than maintaining a natural, full-sounding speech signal, a much more aggressive high-pass of up to 320 Hz can be used. Remember that “telephone-grade audio,” which is optimized for 90% of speech energy, has a frequency response of roughly 300 to 3,000 Hz.
Many microphones have their own high-pass filters. Apply the filter on the microphone first. If more high-pass is needed, activate it on the preamplifier. When high-pass filters are used on both the microphone and preamplifier, the slope of the filter increases, resulting in a quicker falloff of low frequencies.
5. Reduce the Number of Open Microphones
Microphones that are currently recording are sometimes referred to as “open.” When the number of open microphones doubles, the overall background level increases in the mix by 3 dB. For example, when recording a panel discussion with eight participants, there is a 9 dB speech-to-noise penalty when all the microphones are open compared to when only one microphone is open.
Managing a scene with multiple talkers is difficult to do manually, especially for unscripted material. Sound Devices MixAssist and Dugan automatic mixing opens and closes microphones automatically, so sound professionals can focus on the overall mix.
6. Use Real-Time Noise Suppression
Although there are many options for noise reduction via a computer, there have been few portable options for running real-time noise suppression live in the field. CEDAR’s stand-alone noise suppression units have been popular choices for sound professionals who needed portable real-time noise suppression on-location. Now, for the first time in production sound history, adaptive, real-time noise suppression is now available directly in a mixer-recorder with Sound Devices NoiseAssist.
NoiseAssist is a real time, single-ended noise reduction algorithm that uses an advanced algorithm to separate speech from background noise. NoiseAssist is easy to use, with only one setting: the attenuation level of background noise. Up to 20 dB of suppression is available, although 3-6 dB of attenuation is suitable for most situations. Real-time noise suppression tools like NoiseAssist are useful in many applications – for example, fast turnaround productions where there is no time for post, live broadcast and streaming, preview on set, feeds to video village, IFB feeds, rough guide mixes, and more.
When using any kind of noise suppression, record the unprocessed channel for delivery to post-production. Post-production has a whole arsenal of ways to process audio, and having a “raw” track free of any noise suppression is important. The ability to record both a processed and unprocessed track is a powerful feature of 8-Series mixer-recorders and is quite easy to set up.
There are many ways to increase the intelligibility of dialog recordings. Some techniques involve real-world changes like microphone placement, turning off sources of background noise, or using wind screens. Others involve the help of software plugins like MixAssist, Dugan, or NoiseAssist. While NoiseAssist is not a panacea for all recording ills, when used judiciously, it is an indispensable tool for the production sound recordist.