Sound Advice- Audio Delay Line Applications

In Uncategorizedby tfwm

Compared to light and electricity, sound travels at a leisurely 1130 Feet per Second. This relatively slow speed affects many aspects of sound system design and operation. Reverberation as we know it would not exist if sound traveled at the speed of light. Many of the other challenges would be different as well.

In many churches, the problem lies with the auditoriums and venues where a presenter is addressing large groups of people, either unaided or though the use of a central sound system. Audience members may have difficulty hearing or understanding what is being said if they are seated in certain areas of the room. They may be seated in an area that is not covered well by the main speaker system, such as under a balcony. They may also be in a location that is acoustically isolated, such as an overflow seating area. They may also be simply too far away. Hearing can be difficult in these areas because the sound level from the presenter or the main speaker system may be too low compared to the ambient noise or room reverberation.

In most cases auxiliary speakers are added locally to obtain sufficient sound level in these areas. However, without proper consideration for the slow speed of sound, intelligibility can still be a problem. If the auxiliary speaker system is a good distance away from the main speaker system, the auxiliary system will be heard first, followed by the main system, which is heard as an echo. Depending on the delay time and volume, the echo can interfere with listening. Additionally, if the auxiliary speaker is not between the listener and the original sound source, the sound will appear to come from the wrong direction (the auxiliary loudspeaker) and this can distract the listener.

In the 1940s, Helmut Haas researched the disrupting affect delayed sound can have on the listener and how delay affects our localization of sound. Using principles based on his research, and with the use of modern digital delay lines, the echo and localization problems described above can be overcome, and the integration of auxiliary speakers can now be virtually seamless.

Haas’ experiments involved listening tests where one presenter’s voice was reproduced by two speaker systems; one of which using a magnetic audio-tape delay. He observed that if the sound arrives from the delayed speaker between 1 and 30 ms after the original, the delayed speaker is not heard at all, even if the volume from both speakers is the same. The sound appears to come only from the non-delayed speaker. However, the perceived volume will be louder resulting from the combined power of both speakers. He further observed that if one speaker is delayed 5 to 30 ms, the delayed speaker needed to be 10 dB greater in volume than the reference speaker in order for the listener to perceive the volume from the two speakers as equal. As the delay time was further increased, the volume difference must be decreased for the two speakers to be heard at the same level. Although the sound quality changed somewhat with the delay, it was not considered disturbing, and actually made listening less tiring and more natural. As the delay reached approximately 50 ms, it was possible to discriminate the delayed speaker as a separate echo.

Haas further observed, depending on the rate of speech, that if the amplitude of the echo was equal in volume to 10 dB greater in amplitude than the original sound, delays of 40 to 50 ms would disturb only a small percentage of listeners. If the echo signal was reduced in amplitude to 10 dB below the original, no amount of delay disturbed the listeners.

Based on his research, auxiliary loudspeakers benefit from sufficient delay of their signal so that the delayed sound arrives at the listener 5 to 30 ms after the arrival of the original sound. Even as the amplitude of this local speaker is increased, the additional delay will help move the sound image toward the original source. As long as the difference in arrival time within the coverage area of the two speaker systems is 45ms or less, an echo will not be perceived, and listeners should not be disturbed.