I have vibration data measured via isotronic accelerometer and recorded as 16 bit, Fs=22050Hz, little endian binary. (Attached) It was filtered: bandpassed 20Hz to 5000Hz, IIR (Butterworth topology: Order 3). When the attached signal is imported into audacity, I still get a lot of noise (Hiss) in the background. I will like to remove these noise to extract the faint signal (audible via audacity: import). What will be the best filter to use?
It may be worth checking out some of the plugins from Izotope.
http://izotope.com/products/audio/rx/
Is this the kind of thing you're looking for?
Hi Tim,
Not really. I have acquired those signals using virtual instrument, Labview. I am sure there are ways to get rid of those sound by software manipulation. There are various algorithm available to choose from in labview. Some posting suggested using low pass filter to remove the high frequency hiss.
May be FFT of the audio signal will give you an idea about which frequency to supress. Then you can decide on the type of filter to choose.
Try using a logartihmic (rather than linear) scale for your x-axis with a range from 20Hz to 20kHZ. I would also expect the hiss you describe to occupy the frequency range above 2kHz. Try that and repost you findings.
Also remember that the human hearing system is not linear and is most sensitve at 4kHZ http://www.nu9n.com/images/fletcher.jpg
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing