Creating filters with arbitrary response characteristics for use in hearing and speech research

Area of Science:

• Digital signal processing
• Acoustics and audiology
• Biomedical engineering

Background:

• Conventional digital filters (lowpass, highpass, bandpass, band reject) are widely used in speech and hearing research.
• Specialized research applications sometimes require filter characteristics beyond those offered by standard frequency response curves.
• Limitations of conventional filters necessitate the development of more adaptable signal processing tools.

Purpose of the Study:

• To present a straightforward methodology for creating digital filters with arbitrary amplitude and phase responses.
• To offer a versatile alternative to conventional digital filters for specialized signal processing tasks in acoustics and audiology.
• To demonstrate the practical utility of the proposed filter generation technique through sample applications.

Main Methods:

• The core method involves calculating the impulse response of a finite impulse response (FIR) filter.
• The desired filter magnitude and phase response are specified in a text file.
• The calculated impulse response is then convolved with the input signal to achieve the desired filtering effect.

Main Results:

• The described method enables the generation of digital filters with virtually any specified amplitude and phase response.
• This approach provides a high degree of flexibility in filter design, overcoming limitations of conventional filter types.
• Successful application of the method was demonstrated through illustrative examples.

Conclusions:

• The presented technique offers a simple yet powerful way to design custom digital filters for advanced signal processing in speech and hearing.
• This method significantly expands the range of achievable filter characteristics, supporting novel research avenues.
• The convolution-based approach provides a practical solution for implementing highly specific digital filters.