Cochlear Implant Speech Coder Pitch Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current cochlear implants based on the 'channel vocoder' concept struggle with encoding temporal fine structure, leading to poor speech recognition in noise and reduced intelligibility, as they primarily focus on temporal envelopes while discarding fine structure and have inadequate filter density at high frequencies.
Innovation Solution
Implementing a phase vocoder to extract a slowly-varying frequency modulation component and adapting code-excited linear prediction (CELP) vocoders to encode fine structure, which is then added to the cochlear implant processor, improving pitch encoding and overall sound processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If continuous-interleaved-stimulation (CIS) strategies are used to improve temporal envelope representation, then temporal envelope representation is improved, but temporal fine structure is totally discarded
Solution Approach 1:
The patent segments the speech processing into distinct components: temporal envelope extraction and temporal fine structure extraction. By separating these functions, the system can process and preserve both types of information independently, allowing CIS to handle envelope representation while a separate mechanism preserves fine structure for pitch encoding.
Solution Approach 2:
The patent extracts temporal fine structure information separately from the temporal envelope using a phase vocoder. This extraction process isolates the fine structure components that would otherwise be discarded by CIS, enabling their preservation and subsequent use for pitch encoding without interfering with envelope representation.
2Measurement precision
If higher filter density at low frequencies is used to improve fundamental frequency (F0) encoding, then F0 encoding is improved, but filter density at high frequencies is reduced which degrades speech intelligibility
Solution Approach 1:
The patent applies local quality by using a non-uniform filter bank where filter density varies with frequency. Higher filter density is concentrated at low frequencies where pitch information is most critical, while lower filter density at high frequencies preserves speech intelligibility. This localized optimization allows F0 encoding to be improved without sacrificing overall speech quality.
3Measurement precision
If channel vocoder concept is used for cochlear implant processing, then temporal envelope extraction is achieved, but temporal fine structure is lost leading to machine-like sound quality
Solution Approach 1:
The patent merges the traditional channel vocoder approach with phase vocoder technology. The channel vocoder provides robust temporal envelope extraction while the phase vocoder component extracts and preserves temporal fine structure. By combining these two approaches, the system achieves both accurate envelope representation and fine structure preservation, eliminating the machine-like sound quality while maintaining speech intelligibility.
Data Source
AI summary
Cochlear implant performance is improved by extracting pitch information and encoding such pitch information into the processor of a cochlear implant. One embodiment of the invention is to explicitly extract the pitch and deliver it to the cochlear implant by co-varying the stimulated site and rate. Another embodiment of the invention is to implicitly encode the pitch information via a code book that serves as the carrier of stimulation in the cochlear implant.