Title: A%20tutorial%20on%20acoustic%20measurements%20for%20the%20non-technician
1A tutorial on acoustic measurements for the
non-technician
- Svante Granqvist
- Royal Institute of Technology (KTH)
- Dept of Speech Music and Hearing (TMH)
- Stockholm, Sweden
- svante.granqvist_at_speech.kth.se
2Todays topics
- Sound and microphones
- Room acoustics
- Calibration
- Recommendations
3Conclusions
- Use omni-directional electret or condenser
microphones whenever possible - Do not use directed (e.g. cardioid) microphones
unless you really need the directivity - Especially not close to the speaker
- Avoid dynamic microphones
- Place the microphone within the reverberation
radius of the room - Keep noise level low
- Establish a routine for level calibration
4What is sound?
- Demo of sound field
- Sound pressure (pascals, Pa)
- Sound pressure level, SPL (decibels, dB)
- Particle velocity (metres per second, m/s)
- Particle velocity level? Rarely!
5Sound pressure
- Simple relation to sound intensity
- Our ears are mainly pressure sensitive
- Simple relation to distance (1/r)
- Doubled distance gt halved SP ltgt SPL -6dB
- Pressure has no direction
- Pressure sensitive microphones are
omni-directional (no directivity) - So how do they make directed microphones?
6Particle velocity
- Particle velocity has a direction
- So it can be used to create directivity!
- Particle velocity only gt figure of eight
- Mainly sensitive in two directions
7Cardioid
- Particle velocity and sound pressure combined gt
cardioid - Mainly sensitive in one direction
8Directivity
- Omni-directional (SP only)
- Directed (involves particle velocity)
- Figure of eight
- Cardioid
- Super-cardioid
- Other special directivity patterns
- Great!
- ...or is it?
9Directed microphones
- We are primarily interested in sound pressure
- ...but also measure particle velocity
- ...then PV and SP have to be proportional to one
another! - Are they?
10NO!
11Particle velocity
- Particle velocity is proportional to sound
pressure, but only in the far field (1/r) - In the close field, it differs! (1/r2)
- The limit between far and close field depends on
frequency
12Particle velocity
- Particle velocity exhibits a bass lift in the
close field - proximity effect
13Proximity effect (cardioid mic)
14Proximity effect
15Manufacturers data sheets
Cardioid
16Manufacturers data sheets
Cardioid
17Manufacturers data sheets
Cardioid
18Manufacturers data sheets
Omni-directional
19Manufacturers data sheets
- Frequency responses are mostly measured in the
far field, even for microphones that obviously
are intended to be mounted in the close field - You have to add the proximity effect for directed
microphones to those curves!
20Proximity effect (cardioid mic)
21Manufacturers data sheets
Cardioid
22Manufacturers data sheets
Cardioid
23Manufacturers data sheets
Cardioid
24Manufacturers data sheets
Omni-directional
25Demo
- Proximity effect
- Hear the bass lift from the directed microphone
26OK, point taken, he doesnt like directed
microphones
- But then, why are there so many directed
microphones out there? - Music industry, broadcasting, stage use etc
- A bass boost of a few dBs does not matter much or
might even be desired (sound better) - Noise supression may be more important than a
flat frequency response - Most recordings do not have a scientific purpose
27Transducer type
- Electret/condenser
- Can easily be made to have flat response
- Cheap electret microphones (lt 30 ) can be of
sufficient quality - Requires battery/power supply
- Sensitivity may decrease towards end of battery
life - Dynamic
- Difficult to acheive a flat response
- Good dynamic microphones are expensive
- Rarely purely pressure sensitive (even though
data-sheet may say so) - No need for battery/power supply
28Bottom line...
- Use omni-directional, electret/condenser
microphones for scientific purposes! - Make sure batteries are fresh or use some other
type of power supply
29Room acoustics
- In a room sound originates from
- the sound source, directly
- or from reflections at the walls
30Room acoustics
31Room acoustics
- Reverberation radius, rr
- The distance where reflected and direct sound are
equally loud - Less absorbtion gt stronger reflections gt
smaller rr
32Room acoustics
33Room acoustics
- Reverberation radius
- At what distance is the direct sound as loud as
the sound that has been reflected from the walls - Typical value 4 0.5 meters
- Reverberation time
- How long does it take for the sound level to drop
by 60 dB? - Typical value 0.5 4 seconds
34Room acoustics
- How to measure Reverberation time/radius
- Several ways, one would be to record a bang and
see at what rate the sound level drops - The time for a 60dB drop corresponds to
reverberation time - Calculate reverberation radius from this time
35Bottom line...
- Within the reverberation radius, conditions are
similar to free field - Outside, reflections from the walls dominate the
sound - So, put the microphone (well) within the
reverberation radius!
36Level calibration
- Most common method
- Record a signal with a known level
- i.e. a calibration tone
- By relating the level of the calibration tone to
the levels of the signals of interest, absolute
calibration is acheived
37Calibration file, example
Calibration tone
The level was 89 dB
38CalibrationCalibrator
- Procedure
- Mount and start the calibrator (2-10 seconds)
- Unmount calibrator and say the level of the
calibrator - Advantages
- Stable calibration tone
- No sensitivity to room acoustics or surrounding
noise - Disadvantage
- Calibrator that fits the microphone required
- Important that the seal is tight!
39Calibration Loudspeaker SPL meter
- Procedure
- Beep at 1kHz 80 dB (2-10 seconds)
- say the level as read on the level meter
- Advantages
- Stable calibration tone
- Disadvantage
- Loudspeaker signal source reqiured
- Some sensitivity to room and surrounding noise
40Calibration Voice SPL meter
- Procedure
- Sustain /a/ 80 dB (5-10 seconds)
- say the level as read on the level meter
- Advantages
- No loudspeaker required
- Calibration signal (voice) has approximately the
same spectrum as the signals of interest - Disadvantages
- Hard to keep the level of the /a/ stable
- Some sensitivity to room and surrounding noise
41Calibration Voice SPL meter
- Procedure
- Sustain /a/ 80 dB (5-10 seconds)
- say the level as read on the level meter
- Advantages
- No loudspeaker required
- Calibration signal (voice) has approximately the
same spectrum as the signals of interest - Automatic compensation for microphone distance
- Disadvantage
- Hard to keep the level of the /a/ stable
- Only valid for this particular distance
- Some sensitivity to room and surrounding noise
- dB meter should be within rr
42Calibration, directed microphones ?
- Only in the far field (gt30 cm), but still within
rr - Only for rough estimation of SPL
- Never use SPL calibrators!
- Dont try this at home
43Distance compensation
- Sound pressure drops as 1/r
- Re-calculate SPL to appear as recorded at a
different distance, e.g. record at d25 cm, but
report at d130 cm. - Only for omni-directional microphones!
- Formula
44Bottom line...
- Establish a routine for calibrations
- Dont calibrate directed microphones
- Report SPL at 30 cm
- Compensated or actual
- Beware of the mixer on most PC soundcards
45Recommendationsmicrophone and room acoustics
- Depend on
- the purpose of recording
- the recording environment
- Noise
- Room acoustics
Example of purposes SPL Spectrum F0 Inverse
filtering HNR Perceptual evaluation
46SPL
- Omni-directional electret/condenser microphone
- If noisy environment
- Try to attenuate the noise
- Shorten microphone distance (10 cm to the side of
the mouth) - Avoid directed microphones for this purpose!
- Put the microphone well within the reverberation
radius of the room (rr/2) - Re-calculate or calibrate for 30 cm
47Spectral properties (spectrogram)
- Omni-directional electret/condenser microphone
- If noisy environment
- Try to attenuate the noise
- Shorten microphone distance (5-10cm to the side
of the mouth) - If background noise still is a problem a directed
microphone can be used, but beware of the
proximity effect and keep microphone distance
constant! - Put microphone well within rr
48Spectral properties (LTAS, H1-H2, line spectra)
- Omni-directional electret/condenser microphone
- If noisy environment
- Try to attenuate the noise
- Shorten microphone distance (5-10cm to the side
of the mouth) - Do not use a directed microphone
- Put microphone well within rr
- Pay attention to reflective surfaces such as
windows, manuscripts etc.
Added proximity effect, cardioid at 5 cm
49F0, jitter/shimmer
- Any decent microphone is OK, since periodicity is
independent of frequency response - If noisy environment
- Try to attenuate the noise
- Shorten microphone distance (5-10 cm)
- Use a directed microphone
- Check if F0 algorithm is affected by a bass lift!
50Inverse filtering
- Omni-directional electret/condenser microphone
flower lt 10 Hz - Reduce background noise as much as possible
- Never use a directed microphone
- Microphone distance 5-10 cm
- Within rr/10
- Pay attention to reflective surfaces such as
windows, manuscripts etc. - Anechoic chamber is preferred
Addition of reflection to the direct signal
51Harmonics-to-noise ratio
- Omni-directional electret/condenser microphone
- Background noise must be lower than voice noise
- Microphone distance 5-50 cm
- Well within rr
52Perceptual evaluation
- Omni-directional electret/condenser microphone
- Reduce background noise as much as possible
- Microphone distance 5-50 cm
- Well within rr
53But you never know...
- For example, the first intention may be to only
extract F0 - It might turn out, after the recordings are made,
that the recorded material would be suitable for
some other measurement, like SPL - Therefore, do it right from the start!
54Conclusions
- Use omni-directional electret or condenser
microphones whenever possible - Do not use directed (e.g. cardioid) microphones
unless you really need the directivity - Especially not close to the speaker
- Avoid dynamic microphones
- Place the microphone within the reverberation
radius of the room - Keep noise level low
- Establish a routine for level calibration
55These were my recommendations
- You may find reasons to not follow them
- But they better be good... ?
56Questions?
This presentation is available on the
webwww.speech.kth.se/svante/pevoc5 svante.granqv
ist_at_speech.kth.se