Listening With Nature's Gear to Fix Audio

As much as I like test and measurement equipment, sometimes the best troubleshooting tools for a job are two good ears and the gray matter between them.

And maybe a couple of test CDs.

That was the case for one of my consulting projects for a facility that converted a video edit suite into an audio post room with a digital audio workstation.

The room was mid-sized, with the console furniture, DAW, audio mixing board and outboard equipment rack situated on a raised floor. The producers sat below and in front of the audio mixing console furniture, on the main floor level.

A pair of high quality monitor loudspeakers (this was a stereo mixing room for television) was mounted on stands just in front of the front wall, on either side of a video monitor. They were more or less aimed at the audio mixer.

The client's main complaint was that the producers weren't hearing the same sound as the audio mixer, and that what they were hearing was too bass heavy. Also they said it was hard to hear the dialog. The client presented an excellent description of the room, and already I had some idea of what might be going on.

One approach to this project would have been to haul in the TEF analyzer, make measurements, after listening tests, at both the producers' and audio mixer's seats, compare the printouts of the two and confirm that yes indeed the sound was different, and offer some solutions.

But this client had a very limited budget, and was looking for low- or no-cost solutions. A major overhaul of the room wasn't going to be possible at that time.

So for this job I decided to take an experiential approach, doing a variety of listening tests, providing some education, and then experimenting with some practical and inexpensive solutions.

The audio mixing engineer welcomed this approach.

We started by discussing his mixing experiences in the room and any particular issues he faced. As we talked, we walked around the room, and just by listening in different parts of the room, right away it was obvious that the sunken producers' area was a different acoustical space.

LISTENING TESTS

We next moved on to listening tests. We first played some of the pieces that the audio mixer worked on, and listened at the main mix position and at the producers' seating area to hear the sonic differences.

Then we used a couple of test CDs that I particularly like.

The first was the SynAudCon "Test CD for Sound Reinforcement Systems." This disc contains such test signal tracks as 400 Hz and 1,000 Hz tones, full-bandwidth and limited (speech intelligibility range) bandwidth pink noise; pink noise at various octave bands starting with the 31 Hz band up to the 16 kHz band; and individual tracks of pink noise in bursts of one to five seconds at octave intervals from 125 to 8 kHz. Other tracks include shaped tone burst, word lists, speech (lecture), and polarity pulses.

We first listened to the full bandwidth and then to octave band pink noise, one loudspeaker on at a time.

Once we reached the 1 kHz band it was easy to hear, as I suspected, that the producers were quite off-axis from the high frequency drivers of the loudspeakers. As we stood up, we could hear the higher frequency bands becoming louder. The loudspeakers were mounted so that, as is typical, the high frequency driver was above the low frequency driver.

The 2,000 Hz octave band is the important one for good speech intelligibility. Since the level in this band, and the other high frequency bands as well, were lower when listened to off-axis, speech intelligibility was severely impaired.

Also since these levels were lower, the bass frequencies, which roll off less rapidly off-axis than the highs, predominated, and probably even masked some of the highs. So no wonder the producers had trouble hearing the dialog.

The solution was to turn each of the loudspeakers upside down, so that the high frequency driver was on the bottom. Fortunately, this was fairly easy for the particular loudspeakers used, and was done that afternoon so we could re-aim them and re-test.

We aimed the high frequency drivers of the loudspeakers a little lower than ear height for the audio mixer, to provide the producers with a more on-axis sound. After some experimentation and listening, we found a position that was acceptable to both listeners. The audio mixer was happy with this setup, and I later heard that the producers were as well.

SWEET SPOT NOT

While we were conducting the pink noise listening tests, we found that the low frequency response at the audio mixer's position varied with frequency. This was a surprise to the audio mixer, who thought that he had a smooth sweet spot to work in.

We did further investigation with another test CD, the Prosonus SRC Studio Reference Disc. This disc provides individual tracks with single frequency sine waves at 0 and -10 dB, 31 Hz to 15.5 Hz, and also TEF sweeps; full bandwidth pink noise; impulse click; sawtooth pitch reference frequencies; and an 88-note piano scale among other test tracks.

We used the low frequency tracks to listen to some room modes. These occur at frequencies whose wavelengths are comparable to the dimensions of the room. All rooms experience these modes; the smaller the room, the more pronounced.

We found that at some low frequencies the audio mixer's position was in a null (you couldn't hear the tone) and at others it was in a peak (very loud).

Sitting in the audio mixer's chair, we each in turn were able to move back and forth, side to side, to find a spot that was neither in a null or peak for most of the low frequency range and found a good compromise. The audio console furniture would have to be moved back a few inches, but that was doable for the client.

What we discovered, thanks to a facility engineer who had architectural drawings, was that although the room looked rectangular, the outside concrete block walls formed an almost cubic shape, about the worst shape you'd want for a control/monitoring/mixing room. Because of the almost equal dimensions of length, width, and height, the low frequency modes coincide with each other, providing the deep nulls and high peaks.

Other shapes that should be avoided are those where the dimensions are integer multiples of each other.

There were other suggestions presented to the client in order of cost, and I don't know how many were eventually implemented. Those that were done allowed the client to work and hear better in the room that they had, which was their goal.

If they have access to other space in the future, they might want to consider a room designed and built for this purpose, with the proper dimensions, equipment and personnel placement, and acoustical treatment. But in the meantime, they are able to make better television by enhancing the audio.