# US 2019/0244629 A1 · Computer-aided conversion system and method for generating intelligible speech

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

- Number: US 2019/0244629 A1
- Document ID: US20190244629A1
- Status: Published application
- Kind: A1
- Application number: 16/250,645
- Filed: 2019-01-17
- Published: 2019-08-08
- Earliest claimed: 2018-02-05
- Assignees: Ts Voice Technology, LLC
- Applicants: TS Voice Technology, LLC
- Inventors: Shu Wei Tsai; Hengchin Yeh; Tieh-Hung Chuang; Yi-Hsin Chen
- Web page: https://patentgrep.com/patent/US20190244629A1

## CPC

- G10L 21/02: Speech enhancement, e.g. noise reduction or echo cancellation (reducing echo effects in line transmission systems H04B3/20; echo suppression in hands-free telephones H04M9/08)
- G10L 21/013: Adapting to target pitch
- G10L 21/0364: for improving intelligibility
- G10L 25/30: using neural networks
- G10L 15/16: using artificial neural networks
- G10L 2021/0135: {Voice conversion or morphing}
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING

## Abstract

The present invention provides a computer-aided conversion system and method for generating intelligible speech that uses a transmitter disposed in the nasal cavity of a user and a receiver disposed in pairs with the transmitter, the transmitter transmits a detecting signal in waveform to the nasal cavity of the user, and the receiver receives a reflected wave from the user's nasal cavity. After analyzing the reflected wave, a spectrum corresponding to the acoustic model of an articulatory cavity is obtained. Through the spectrum, the intention in the speaking of the user may be known, that is, the present invention may detect a speech not originated from the vocal cord of the user.

## Claims

### Claim 1

A computer-aided conversion system for generating intelligible speech used for obtaining a speech not originated from a vocal cord of a user, the computer-aided conversion system for generating intelligible speech comprising:
a detection unit, comprising a transmitter disposed corresponding to the nasal cavity of the user, and a receiver disposed in pairs with the transmitter; and
an computation unit, comprising a transmitting-computing module to generate a detecting signal and transmit the detecting signal to the transmitter, and a receiving-computing module to analyze a reflected signal received by the receiver, wherein the transmitter transmits the detecting signal in a waveform to the nasal cavity of the user, the receiver receives a reflected wave originating from the nasal cavity of the user and converts the reflected wave into the reflected signal, the receiving-computing module obtains a spectrum corresponding to an acoustic model of an articulatory cavity based on the reflected signal, and the spectrum corresponds to a speech not originated from the vocal cord of the user.

### Claim 2 (depends on 1)

The computer-aided conversion system for generating intelligible speech as claimed in claim 1, wherein the transmitter is disposed in a nasal-oro-pharyngo-laryngeal tract of the user, and the receiver is disposed at a position within 3 cm in front of lips of the user.

### Claim 3 (depends on 1)

The computer-aided conversion system for generating intelligible speech as claimed in claim 1, wherein the detection unit comprises a cushion disposed on the transmitter.

### Claim 4 (depends on 1)

The computer-aided conversion system for generating intelligible speech as claimed in claim 1, wherein the detecting signal is configured as an ultrasonic wave of 16 to 22 KHz.

### Claim 5 (depends on 1)

The computer-aided conversion system for generating intelligible speech as claimed in claim 1, further comprising a representation unit, the representation unit comprising a database storing acoustic information corresponding to the spectrum, and a player audibly or visually indicating a speech not originated from the vocal cord of the user.

### Claim 6

A computer-aided conversion method for generating intelligible speech used for obtaining a speech not originated from a vocal cord of a user, the method comprising following steps of:
step S1: transmitting a detecting signal to the nasal cavity of the user, wherein the detecting signal is transmitted in a waveform;
step S2: receiving a reflected wave of the detecting signal reflected in the nasal cavity of the user, and defining the reflected wave as a reflected signal;
step S3: analyzing the reflected signal to obtain a spectrum corresponding to an acoustic model of an articulatory cavity; and
step S4: obtaining a speech not originated from the vocal cord of the user based on the spectrum.

### Claim 7 (depends on 6)

The computer-aided conversion method for generating intelligible speech as claimed in claim 6, wherein the reflected signal is converted into the spectrum by Fourier transform.

### Claim 8 (depends on 7)

The computer-aided conversion method for generating intelligible speech as claimed in claim 7, wherein the step S3 comprises the following steps of:
step S31: intermittently sampling the reflected signal after the Fourier transform;
step S32: executing an adaptive controller to obtain a characteristic information based on the reflected signal after the Fourier transform; and
step S33: generating the spectrum based on the characteristic information of the reflected signal after the Fourier transform.

### Claim 9 (depends on 8)

The computer-aided conversion method for generating intelligible speech as claimed in claim 8, wherein the adaptive controller is an artificial neural network, and the characteristic information is a Mel-frequency cepstrum obtained after the reflected signal transformed by the Fourier transform is mapped to a Mel scale.

### Claim 10 (depends on 6)

The computer-aided conversion method for generating intelligible speech as claimed in claim 6, wherein before the step S 1, further comprising the following steps of:
step S01: waiting for a start command;
step S02: searching for a calibration parameter corresponding to the user based on the start command; and
step S03: determining the detecting signal based on the calibration parameter.
