Conclusion
Summary: Conclusion of determining room inpulse responses and deconvolving the response out of a recorded signal.
Conclusion
After taking the impulse responses and comparing them to the theoretical values, we noticed a large difference due to non-rectangular rooms, objects in the room and clipping. The non-linear effect of clipping not only removed information from the signal, but did so in a way that was unrecoverable by our Fourier analysis. With a better theoretical model, we could take responses from multiple sources and recievers quickly, but for getting a single impulse response of a room, given little to no clipping, manually taking the impulse response is simple enough. For commercial uses of measuring the impulse response, like measuring the response from each seat in a orchestra hall to each instrument, a robust theoretical method is needed, including other objects in the room and non-rectangular rooms. Materials played an important factor in the impusle response. While most of the walls had similiar reflection coefficients, the Will Rice room had wooden ceilings that had a drastic impact on the impulse response. The lower reflection coefficient led to more distortion in the signal and a loss of energy compared to the other surfaces.
The deconvolution of the signal was intended to remove the room response on a recorded signal, but in the process amplified the noise. Much of the noise was in the signal and could not be easily filtered out. The deconvolution did reproduce the original signal, the quality was worse than the recorded signal. With a better method of deconvolution, one that was able to account for noise and minimize it rather than amplify it, clearer signals could be produced. The clipping of our signal caused the deconvolution to remove part of the response that was already taken out, creating amplitude aliasing. Perfect deconvolution would be useful in creating a clean recorded signal, almost regardless of recording environment. Deconvolution has uses in non-sound signals as well. Given a signal and the impulse response of the environment in which it was taken, the original signal should be retrievable through deconvolution given the system is linear and time-invariant. For specific cases, a true impulse response isn't needed. If the response for all possible frequencies is known, it can be used instead of the impusle response with the other frequencies filtered out to remove noise. The specific response can be deconvolved from a known input signal and the recorded output of the system.
Comments, questions, feedback, criticisms?
Send feedback
- E-mail the author of the module, Conclusion
Footer
More about this content: Metadata | Version History
This work is licensed by David Newell under a Creative Commons Attribution License (CC-BY 2.0).
Last edited by David Newell on Dec 13, 2005 4:58 pm US/Central.