160 likes | 364 Views
Title: A practical comparison of the ”new” swept-sine technique and MLS in measuring room impulse responses and acoustical parameters. Lars H. Morset Morset Sound Development Trondheim, Norway.
E N D
Title: A practical comparison of the ”new” swept-sine technique and MLS in measuring room impulse responses and acoustical parameters Lars H. Morset Morset Sound Development Trondheim, Norway
Studied acoustics at NTNU (prof. Krokstad, now prof. P. Svensson), now doing Ph.D on violin acoustics Started Morset Sound Dev. 1997, developing WinMLS – a sound-card based software for audio, acoustical & vibrational measurements. Old MLS: Maximum Length Sequence New MLS: Measurements based on Log. Sine-sweep My background
Assumes our system is linear and time-invariant (LTI) linear: a*x + b*y = if we double the input signal to the system, the output signal is also doubled if we superpose two signals time-invariant: system does not change as function of time the impulse response completely describes a LTI system Background Theory
No, a loudspeaker is not completely linear it will have some distortion (harmonic), increasing with output level a room is not completely time-invariant the air will move temperature changes time-variance usually a bigger problem for large rooms Are our conditions really LTI?
Harmonic distortion cannot be separated from the impulse response the loudspeaker is never completely linear harmonic distortion will happen and it is increases with the output level, SPURIOUS PEAKS Solution: Do not use too high output level but a high output level is desired... Want maximum SNR Size/cost of loudspeaker MLS and linearity
MLS is not robust against time-variance rooms are never time-invariant usually a problem for high frequencies longer measurement time does not increase SNR when we have time-variance Sampling frequency input/output has to be exactly the same to avoid loss of quality not a problem for “clocked” hardware always a problem for remote measurements MLS and time-variance
MLS “remote” loop-back measurement (pre-filtering used)Problems with high-frequency
Loudspeaker non-linearity can be kept to a minimum by using high quality elements and avoiding using high output levels Time-variance in rooms turn off ventilation make sure people are not moving around during measurements avoid temperature gradients Perform remote measurements with device that has the same clock How do we deal with this in general?
It has to be post-processed correctly! Not the same as traditional swept sine A higher loudspeaker output levels can be applied Robustness to minor time-variance. Make sure a single sweep only is used Remote measurements can be performed without loss in quality Make sure a single sweep only is used A logarithmic swept sine “solve” these problems
Seems to still be a problem measuring large halls (London conference on auditorium acoustics-2002) We now have the solution! Do not want cables on the floor (remote) Multi-channel measurements (24 channels) Very good SNR obtained with 5-20 seconds log-sine sweep The log-sweep is not that bad to listen to Occupied measurements
In the general case of measuring room impulse responses and acoustical parameters, a long (5-40 seconds), single sweep should be applied removes uncertainties with harmonic distortion and time-variance are left “only” with transducer uncertainties and impulse response post-processing For measuring sound systems in small rooms (studios), the sweep length can be as low as 1 sec. In this application, MLS can also successfully be used What is today's “best” measurement technique?
You need this hardware: PC with soundcard (usually built-in) Microphone Loudspeaker Evaluate WinMLS in 30 days for FREE by downloading from www.winmls.com Inexpensive versions for personal usage Have information/CDs available here How can I myself test these methods?