[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Measuring Reverberation time
Dear Ben and List,
I have measured reverberation time for a speech intelligibility study in
small rooms. I did not follow any particular standard, but there is a
dense mathematical coverage of room acoustics in Allan Pierce's
'Acoustics', chapter six which I used as a reference. It is not light
Overall I would say there are two considerations with regard to
reverberation time of a given room. One being the location and
orientation of sources and receivers. The second being the frequency
content of the signal. As you might expect these two are interlinked
(directional effects and room modes).
For the reverberation time measurements, I would recommend putting the
source and receiver where you intend to place the speaker and listener
during the intelligibility experiments. I did some preliminary
experiments to determine if listening in a corner or broadcasting from a
corner lower speech intelligibility. It appears that putting a listener
in a corner of a room will reduce speech intelligibility scores. I was
unable to show the same for putting the source in a corner. More work
would need to be done to make a definite conclusion in either case.
Critical radius, Pierce pp. 268, might be useful. It estimates where the
field is dominated by either direct or reflected sound. However it isn't
based on intelligibility, I believe Peutz defined a critical distance
based on intelligibility.
Reverberation time is usually measured in bands, the question is what
bands for speech? If you get any references not posted to the list I
would appreciate a forward. I think Nabelek chose three bands in her work
on speech intelligibility during the 70's and 80's if you want more
information on this I'll look it up. I suspect noise weighted like speech
would also be appropriate. In any case I recommend getting rid of most
noise below 50 Hz or so. It can have an effect on reverberation time, and
little on intelligibility.
One trouble you might encounter is that 60 dB is a fairly large range to
measure. You will probably be measuring noise before any real signal
reaches .0001% of its original energy. To alleviate this, Wilson
recommends extrapolating your decay curve. I used a linear fit between -5
dB and -35 dB.
In my experiments I wanted the impulse response of the room as well as the
reverberation time. I used a broad band source, a starter pistol (a
spark, or popped balloon might work) to generate a pulse. This was
measured both in an anechoic environment and in the room. (remember a
human body will affect the measurements) By convolving these two
responses together I came up with an approximate room impulse response.
I was then able to work with this response digitally to determine the
reverberation time for different bands. This method is a hassle but does
give a complete picture.
Peutz, V. M. A. (1971). "Articulation loss of consonants as a criterion
for speech transmission in a room." J. Audio Eng. Soc., 19: 915-919.
Pierce, Allen D. (1991). Acoustics, and Introduction to Its Physical
Principles and Applications, Woodbury, NY: Acoustical Soc. of
America. Reference to Chapter Six "Room Acoustics" pp.250-312
Wilson, Charles E. (1994). Noise Control: Measurement, Analysis, and
Control of Sound and Vibration, Malabar, FL: Krieger. page 65.
On Thu, 29 Nov 2001, Ben Hornsby wrote:
> Hello All,
> I am looking for references for appropriate techniques for measuring
> reverberation time of a medium sized room (approx. 40'x40'x20'). In
> specific, number and position of loudspeakers, level and type of
> signal. Does your technique and stimuli vary depending on your
> question of interest (e.g. absorption characteristics versus effects
> on sound quality or speech intelligibility)? We are interested in
> speech intelligibility by the way. I have ordered an IEC standard that
> supposedly describes a "standard" method for measuring reverb times
> when examining the absorption characteristics of materials. Any
> thoughts are appreciated.
> Ben Hornsby