| Summary: | 碩士 === 中原大學 === 電機工程研究所 === 99 === This thesis discusses the cancellation of a low frequency snore signal by using ANC based method in a DSP. Generally speaking, snoring is a high up problem in nowadays. The volume of severe snoring can reach up to 90 dB, which not only cause hearing loss but also disturb the sleep of bed partner. It also linked to extremely sleepiness in daytime of snorer and his bed partner, leading to the loss of productivity in the work, safety of driving and occupation. Most solutions to this problem use passive earmuffs or earplug for the bed partner. But these facilities are uncomfortable to be wear when sleep. Another method like anti-snore pillow or tooth socket for the snorer is uncomfortable, either. A period of snore can be divided into two components, inspiration and expiration. The frequency of snore is also in low frequency, between 100 to 300 Hz. Accordingly based on the theory of superposition, the ANC system is a better way to reduce the snore noise by generating an artificial anti-snore signal
This thesis uses the popular filtered-X least mean square (FXLMS) method in ANC to derive the proper signal to cancel the undesired snore noise. By using a DSP chip, this method can be realized through the estimation of secondary path, from the location of secondary speaker to the error microphone, and then tune the weights of the FIR filter to generate the anti-snore noise. This artificial noise has the same magnitude and 180 degrees phase shift of the annoying snore. Therefore, the anti-snore noises can interference the snore signal, producing a silent zone at the ears of bed partner.
|
|---|
