What is the frequency response in microphones

Sound design - microphones with extended frequency response

by Klaus Baetz,

If you look at the frequency response of various condenser microphones, you will usually find an indication of 20 Hz – 20 kHz. In other words, these microphones can pick up noises and tones within this frequency range with relative ease. In principle, that sounds like a sensible dimensioning, because a young, healthy person can also perceive noises and tones in this area. In this sound design episode, we want to think outside the box and deal with what happens outside of these limits, preferably in higher hertz ranges.

Before we turn to the technical details, let's first ask the most important question: Why all this? If we can only hear up to a maximum of 20 kHz, what do we get out of it? A legitimate question that is easy to answer. Firstly, the higher resolution allows for a higher degree of precision in pretty much all work steps in post-production. The oversampling, which is increasingly used nowadays, at least partially compensates for this point. That is why we come to the second point, which is more important for sound design: a tonally significantly better pitching. Anyone who has ever pitched down a recording knows the effect that the material unfortunately sounds very dull at some point with higher settings. Here you can generate additional height information with all kinds of tricks such as exciters and distortions, but this only helps to a certain extent. With an extended frequency range we can avoid this problem, because we now "push" the previously inaudible frequency material into the audible range and thus have additional information in the highs, which makes the recording sound much clearer and more natural. In addition to the technical basics, which I will go into in the following, it is of course also necessary that the sound source is actually active in the range beyond 20 kHz.

Microphones

A small addition to the above frequency specification of microphones: It is not the case that these microphones can only pick up the range between 20 Hz and 20 kHz, but the microphone can also process information below and above it. However, the performance of the devices drops here because they are not intended for this purpose. This is where microphones come into play that have an extended frequency range. One example is the Sennheiser MKH-8000 series, which is probably the best-known representative of this guild. Depending on the microphone, the frequency range extends from 10 Hz to 70 kHz. But other manufacturers also offer corresponding microphones, e.g. B. the Japanese manufacturer Sanken with the CO-100K, which goes up to 100 kHz, or the American manufacturer Earthworks, which also offers a very wide range of microphones.

And recording

So that we can use the extended frequency range of the mentioned microphones, a sample frequency of 44.1 kHz or 48 kHz is no longer sufficient, because we are only scratching the surface of the possibilities. According to the Nyquist-Shannon sampling theorem, we need at least twice the sample rate of the maximum frequency that we want to record. That means we have to work with sample rates like 88.2 kHz, 96 kHz or even 192 kHz, depending on what our microphone can do.

Almost all audio interfaces nowadays support sample rates of up to 96 kHz, and many converters now even manage 192 kHz in inexpensive audio interfaces. So the chances are that most of you will be able to take appropriate recordings. Even in the field of portable recorders, almost every current model actually manages at least 96 kHz. Of course, the higher the sample rate, the greater the storage requirements for the audio files and the load on the audio system during editing and processing. But even with this there should be no more problems in most cases nowadays. In short: In principle, there are hardly any obstacles in our way on the recording side.

A first test

To clarify the whole thing a little more, I made a test recording with an Oktava MK 012 and a Sennheiser MKH 8020 at 192 kHz sample rate. This is not about a comparison of the microphones per se, because the Sennheiser microphone not only costs a multiple of the Oktava, it also has a different directional characteristic (omnidirectional versus cardioid). My aim is to use the spectrogram in iZotope RX 6 to show what additional information the MKH 8020 (frequency response: 10 Hz-70 kHz) can capture compared to the widely used Oktava MK 012 (frequency response: 20 Hz - 20 kHz) .

If you compare the two spectrograms with each other, you can see that the MKH 8020 has a significantly expanded frequency range compared to the MK 012. At the same time, however, you can also see that the MK 012 is far from over at around 20 kHz and that this microphone can also benefit from a higher sample rate.

Pitching or Pseudo-Resampling?

Before we get into actual sound design, there is one final important point that may be worth experimenting with. As already mentioned, one of the great charms of microphones with an extended frequency range is that the recordings can be pitched down wonderfully and yet they do not sound dull. However, besides pitching, there is a second method that can achieve the same effect. This is pseudo resampling, i.e. changing the sample rate, but without actually performing resampling. This option is not necessarily supported by all software or can be a little hidden. Pseudo resampling is possible with iZotope RX, for example.

As soon as "Change tag only" is activated, the RX does not resample, but only changes the file properties. Adobe Audition also offers a corresponding option under the menu item "Interpret sampling rate". In Steinberg WaveLab, the setting can be made to the right below the waveform by clicking on the sample rate display, and the Cubase resampling dialog does not seem to do any real resampling either.

The main attraction of this method is that the sample rate is only changed in the file properties. So if a 192 kHz file is converted into a 96 kHz file using this method, it is actually still a 192 kHz file, because the material remains completely untouched. From now on, the audio software will interpret the material as 96 kHz material, which means that the file is played back more slowly and is therefore pitched down, because it contains twice as much information as a real 96 kHz file.

Which method you prefer is of course up to you, and the whole thing depends heavily on the pitching or resampling algorithms available.

In the next month we will dedicate ourselves directly to the practice and see what we can do with our recordings.



PART 2

Our first experiment: We want to make a small and quiet splash of water bigger. To do this, we stir and splash around in a basin with a spoon. Then it should sound as if the splashing was in a cave.

I filled the sink in the kitchen with water just below the overflow, positioned the Sennheiser MKH 8020 approx. 30 cm above the water surface and then stirred the water with a spoon. A little caution is of course required so that the microphone does not get wet, but that should go without saying.

The actual recording took place with a sample rate of 192 kHz in order to be able to use the full resolution of the microphone. After a short post-processing in iZotope RX (more on that later) I "pseudo-resampled" the recording to 72 kHz in order to then perform a real resampling to 44.1 kHz. Then the recording ended up in Cubase, where I gently limited the very dynamic signal via event volume and brickwall limiter.

The last step was the design of a room with Steinberg's Revelation Hall. I started with a church preset, significantly increased the pre-delay and reverberation time and reduced the density so that individual reflections can be heard and the modulation of the reverberation slightly increased to make it more lively.

Due to the strong down-pitching of the signal, it no longer sounds like a spoon in a sink, but rather tends towards "rowing your arms in a body of water" or, with the splashing sounds, towards small stones that are thrown into the water.

A forest of vegetables

The next sound sample came about more or less by chance because it was my turn to prepare lunch and cut a lot of leeks and spring onions. Of course, vegetables like this can also be torn apart before cutting, which is why I quickly grabbed my Zoom H6 and recorded the process. Since the zoom recorder only supports recordings up to a maximum of 96 kHz, I had to be content with this sampling rate for the example, but that is not such a big problem because I did not have to pitch so extremely here. So after recording I pseudo-resampled the recording in RX from 96 to 48 kHz and thus pitched it down an octave. This means that the cracking and tearing of the small vegetables sounds much more like large plants.

Then I inserted the recording back into Cubase and processed it with a multiband envelope shaper. With this I mainly shortened the release times, the more the lower the frequency band was. On the other hand, I slightly emphasized the attacks in the upper two frequency bands. The signal was then sent through the Quadrafuzz 2, whereby the two lower frequency bands were deactivated and a certain roughness and sharpness was ensured in the two upper bands with the tube distortion.

Finally, the signal went through an instance of the Frequency EQ, with the help of which the bass and lower mids were raised in the range of around 250 Hz, so that the overall picture gets a little more fullness.

Interim conclusion

After our first two tests, I would like to point out two problems that I have not yet discussed in this way. The first problem is noise, which occurs with very large pitch processes (from around two octaves). Suddenly a perceptible noise appears in the high-frequency range, which at first seems surprising, since the Sennheiser MKH microphones are really extremely noise-free. Fortunately, this problem can be eliminated with a denoiser à la iZotope RX without much effort.

The second problem concerns the natural reverberation of the room in which the recording was made. The pitching process naturally also makes it deeper and longer, which means that a normal living space can sometimes be transformed into a nave in extreme cases. The directional characteristic of the microphone also plays a role here, because the MKH 8020, which I used for my recordings, has an omnidirectional characteristic and thus picks up sound equally strong from all directions. This can be counteracted with a directional microphone - within the MKH series, however, this also changes the frequency response. For example, the MKH 8040 with cardioid characteristic has a frequency response of 30–50000 Hz compared to 10–60,000 Hz for the MKH 8020, which should be sufficient in most cases. So if you should have the opportunity to use several of these microphones with different directional characteristics, you can decide flexibly which trade-off you want to enter - less space or a slightly lower resolution.

Little swords

We stay in the kitchen and turn to the knife block. Here I took a Santoku knife and pulled it with the broad side over the back of a normal cutlery knife. It was recorded in 192 kHz, the whole thing was then "pseudo-resampled" to 48 kHz and re-imported into Cubase.

The sound already goes in the direction of "drawing a little sword", but it still lacks a little aggressiveness and intensity. Therefore, I first boosted both the attack and the release phase with a multiband envelope shaper in the upper three frequency bands, whereby I also extended the attack phase towards the upper frequencies. This emphasizes and lengthens the actual peak of the "sword-drawing".

Then the signal was compressed even further with the Steinberg Vintage, and finally Steinberg's Frequency EQ followed, which removed a lot of rumble in the bass and boosted the highs slightly and the lower mids strongly. The sound for the next sword duel is ready.

Drones & Soundscapes

Our last example is a shot of the kettle bubbling away. I also recorded this process at 192 kHz, “pseudo-resampled” to 48 kHz and then imported the recording into Cubase on two tracks, shifting one by around 500 ms against the other. The sonic result is a deep, thumping roar and bubbling. Because I panned both tracks hard left and right, a very broad background atmosphere was created, which was then slightly compressed and slightly trimmed in the depths with the frequency EQ. In addition, there was an annoying resonance at around 100 Hz, which was gently lowered. For the finishing touches, the QuadraFuzz 2 was used again to distort the range between 2.5 and 9.7 kHz.

Other possible uses

Of course, the above examples only scratch the surface of the things that are possible if you can make everyday sounds appear much larger and more powerful without much loss of quality through pitching. Such microphones are wonderfully suitable, for example, for voice effects in which the protagonist should speak as deeply as possible and yet easily understandable.

Have fun experimenting!

 

You might be interested in that too