# US 2016/0235987 A1 · Method for Extracting Temporal Features from Spike-Like Signals

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2016/0235987 A1
- Document ID: US20160235987A1
- Status: Published application
- Kind: A1
- Application number: 15/027,307
- Filed: 2014-10-06
- Published: 2016-08-18
- Earliest claimed: 2013-10-07
- Assignees: Technische Universität München; MED-EL Elektromedizinische Geraete GmbH
- Applicants: Med-El Elektromedizinische Geraete GmbH
- Inventors: Michael Isik; Werner Hemmert
- Web page: https://patentgrep.com/patent/US20160235987A1

## CPC

- A61N 1/36038: {Cochlear stimulation}
- A61N: ELECTROTHERAPY; MAGNETOTHERAPY; RADIATION THERAPY; ULTRASOUND THERAPY (measurement of bioelectric currents A61B; surgical instruments, devices or methods for transferring non-mechanical forms of energy to or from the body A61B18/00; anaesthetic apparatus in general A61M; incandescent lamps H01K; infrared radiators for heating H05B)

## Abstract

A system and method for the evaluation of hearing implant signal processing is described. A cochlear implant signal processor converts a speech signal input into multi-channel stimulation pulse sequences for a cochlear implant electrode array. A temporal feature processing module extracts temporal feature information from the stimulation pulse sequences. A speech recognition engine evaluates the temporal information using automatic speech recognition to produce speech recognition outputs corresponding to the speech signal input.

## Claims

### Claim 1

A system for cochlear implant signal processing, the system comprising:
a cochlear implant signal processor configured to convert a speech signal input into multi-channel stimulation pulse sequences for a cochlear implant electrode array;
a temporal feature processing module configured to extract temporal feature information from the stimulation pulse sequences; and
a speech recognition engine configured to evaluate the temporal feature information using automatic speech recognition to produce speech recognition outputs corresponding to the speech signal input.

### Claim 2 (depends on 1)

The system according to claim 1, wherein the temporal feature processing module includes a temporal feature extraction module configured to extract the temporal feature information using a short term spectrum calculation.

### Claim 3 (depends on 2)

The system according to claim 2, wherein the temporal feature processing module includes a temporal information limiting module configured to limit temporal feature information represented by a short term spectrum by discarding phase information and preserving magnitude information.

### Claim 4 (depends on 1)

The system according to claim 1, wherein the temporal feature processing module includes a temporal information limiting module configured to remove frequency content greater than some given channel maximum for each signal channel to extract the temporal feature information.

### Claim 5 (depends on 3)

The system according to claim 3, wherein the temporal information limiting module uses a low pass brick-wall filter to remove frequency content.

### Claim 6 (depends on 4)

The system according to claim 4, wherein the temporal information limiting module uses a band pass filter with the brick-wall filter.

### Claim 7 (depends on 3)

The system according to claim 3, wherein the temporal information limiting module uses a low pass filter to remove frequency content.

### Claim 8 (depends on 1)

The system according to claim 1, wherein the temporal feature processing module includes a temporal information aggregation module configured to aggregate temporal information across signal channels when extracting the temporal feature information.

### Claim 9 (depends on 1)

The system according to claim 1, wherein the temporal feature processing module includes decorrelation module configured to transform the temporal feature information into a different physiological space.

### Claim 10 (depends on 1)

The system according to claim 1, wherein the temporal feature processing module includes a channel limiting module configured to output a set of n-most significant coefficients for speech recognition.

### Claim 11

A method for the evaluation of cochlear implant signal processing, the method comprising:
converting a speech signal input into multi-channel stimulation pulse sequences for a cochlear implant electrode array;
extracting temporal feature information from the stimulation pulse sequences; and
evaluating the temporal feature information to produce speech recognition outputs corresponding to the speech signal input.

### Claim 12 (depends on 11)

The method according to claim 11, wherein extracting the temporal feature information includes using a short term spectrum calculation.

### Claim 13 (depends on 12)

The method according to claim 12, wherein the temporal feature information in the short term spectrum is limited by discarding phase information and preserving magnitude information.

### Claim 14 (depends on 11)

The method according to claim 11, wherein extracting the temporal feature information includes removing frequency content greater than some given channel maximum for each signal channel.

### Claim 15 (depends on 14)

The method according to claim 14, wherein a low pass brick-wall filter is used for removing frequency content.

### Claim 16 (depends on 15)

The method according to claim 15, wherein a band pass filter is used with the brick-wall filter.

### Claim 17 (depends on 14)

The method according to claim 14, wherein a low pass filter is used for removing frequency content.

### Claim 18 (depends on 11)

The method according to claim 11, wherein extracting the temporal feature information includes aggregating temporal information across signal channels.

### Claim 19 (depends on 11)

The method according to claim 11, wherein extracting the temporal feature information includes decorrelating the temporal feature information based on a transform into a different physiological space.

### Claim 20 (depends on 11)

The method according to claim 11, wherein extracting the temporal feature information includes performing channel limiting to output a set of n-most significant coefficients for speech recognition.
