Speech information transmission method and system
First Claim
1. An apparatus for transmitting speech information through a Telex system, comprising:
- (a) speech input conversion means for converting an analog speech signal to be transmitted into a corresponding digital speech signal;
(b) speech extracting and compressing means for preparing a character code train of a low bit rate by extracting at least one of syllable units and phonetic symbols from said digital speech signal and compressing the extracted syllable units or phonetic symbols, and for determining accent types of words or phrases contained in said character code train to produce an accent identifying code;
(c) Telex coding means for converting the character code train of the low bit rate and said accent identifying code into a Telex code signal;
(d) channel means for transmitting the Telex code signal;
(e) code conversion means for converting said transmitted telex code signal into a corresponding signal train of the syllable units or phonetic symbols and an accent type identifying signal; and
(f) speech information decoding means for decoding said signal train and said accent type identifying signal from the output of said code conversion means to synthesize voice speech to thereby reproduce the information of the speech signal,wherein said extracting and compressing means (b) includes a circuit for extracting partial autocorrelation coefficients (PARCOR) and pitch frequency for the input speech signal, a DP matching circuit for determining phoneme information on the basis of the PARCOR signal, and an operation controller circuit adapted to execute processing for converting said phoneme information signal into a syllable series signal and determining the accent type from the pitch frequency series, said code conversion means (e) includes a controller for identifying the character code train from the incoming telex code signal transmitted through said channel and converting the character code train into a time series signal of the syllable units or phonetic symbols, and said speech information decoding means (f) includes a pitch pattern memory for previously storing standard pitch patterns for speech characters and producing corresponding pitch pattern information in response to the incoming character code train, a speech synthesizer including a speech element memory for previously storing predetermined waveforms of speech elements and adapted to respond to said time series signal from said controller means to thereby produce a speech signal of the syllable units or phonetic symbols in the form of a time series of the speech elements each of which corresponds to a pitch unit, whereby said speech element memory is controlled by said pitch pattern information output from said pitch pattern memory to produce a speech waveform signal having a duration determined by said pitch pattern information, and means for converting the speech waveform signal output from said speech synthesizer to a speech signal.
0 Assignments
0 Petitions
Accused Products
Abstract
Speech information communicating method and apparatus for transmitting speech information through a transmission system having a low bit rate transmission capacity such as Telex to a receiver station for reproduction. The apparatus comprises at the side of transmitter station a speech recognition device for extracting by high data compression a character train corresponding to syllable units or phonetic symbol from an input speech and simultaneously determining accent types of words or phrases contained in a character train, and a code converter for encoding the extracted character train and the identified accent type in the form of Telex code, unoccupied Telex codes being allotted to the accent type codes and for producing an encoded speech information signal including a time series of character codes having the accent type identifying codes inserted therein. The receiver station includes a decoder for decoding the encoded speech information signal from a transmission channel into the character code train and the accent identifying code, a signal processing circuit for converting the derived character code train into a syllable-based time series signal while preparing accent type identifying signal from the accent identifying code, thereby to produce a speech synthesis control signal, a speech synthesizer for reading out voice waveform signals to be synthesized from a voice waveform storing memory in accordance with the speech synthesizing control signal, and converter for converting the readout voice waveform signal into a corresponding audible speech signal. The output from the encoding circuit of the transmitter can be once recorded on a Telex tape and transmitted to the receiver later on by an automatic Telex sender system. At the receiver station, the received speech code information can be recorded and reproduced later on by a speech decoding apparatus provided with a tape reader having a speech restoring function.
70 Citations
7 Claims
-
1. An apparatus for transmitting speech information through a Telex system, comprising:
-
(a) speech input conversion means for converting an analog speech signal to be transmitted into a corresponding digital speech signal; (b) speech extracting and compressing means for preparing a character code train of a low bit rate by extracting at least one of syllable units and phonetic symbols from said digital speech signal and compressing the extracted syllable units or phonetic symbols, and for determining accent types of words or phrases contained in said character code train to produce an accent identifying code; (c) Telex coding means for converting the character code train of the low bit rate and said accent identifying code into a Telex code signal; (d) channel means for transmitting the Telex code signal; (e) code conversion means for converting said transmitted telex code signal into a corresponding signal train of the syllable units or phonetic symbols and an accent type identifying signal; and (f) speech information decoding means for decoding said signal train and said accent type identifying signal from the output of said code conversion means to synthesize voice speech to thereby reproduce the information of the speech signal, wherein said extracting and compressing means (b) includes a circuit for extracting partial autocorrelation coefficients (PARCOR) and pitch frequency for the input speech signal, a DP matching circuit for determining phoneme information on the basis of the PARCOR signal, and an operation controller circuit adapted to execute processing for converting said phoneme information signal into a syllable series signal and determining the accent type from the pitch frequency series, said code conversion means (e) includes a controller for identifying the character code train from the incoming telex code signal transmitted through said channel and converting the character code train into a time series signal of the syllable units or phonetic symbols, and said speech information decoding means (f) includes a pitch pattern memory for previously storing standard pitch patterns for speech characters and producing corresponding pitch pattern information in response to the incoming character code train, a speech synthesizer including a speech element memory for previously storing predetermined waveforms of speech elements and adapted to respond to said time series signal from said controller means to thereby produce a speech signal of the syllable units or phonetic symbols in the form of a time series of the speech elements each of which corresponds to a pitch unit, whereby said speech element memory is controlled by said pitch pattern information output from said pitch pattern memory to produce a speech waveform signal having a duration determined by said pitch pattern information, and means for converting the speech waveform signal output from said speech synthesizer to a speech signal. - View Dependent Claims (2, 3, 4, 5)
-
-
6. An apparatus for encoding speech information comprising:
-
(a) A/D converter means for converting an analog input speech signal into a corresponding digital speech signal; (b) an extractor circuit for extracting partial autocorrelation coefficient (PARCOR) data and pitch frequencies from the digital speech signal; (c) a dynamic programming matching circuit for comparing on the basis of said PARCOR data the input speech signal successively with standard patterns previously stored in a standard pattern memory to detect a vowel-consonant-vowel (VCV) train which coincides with the standard patterns, to thereby segment said input speech signal into syllable units; (d) an operation control circuit for converting the syllable units of the segmented input speech signal into character codes and estimating accent position in a syllable unit series of the input speech signal on the basis of said pitch frequency series to thereby produce corresponding accent identifying codes; and (e) means for combining said character codes with Telex code positions in one-to-one correspondence while combining said accent identifying codes with telex unallocated code positions.
-
-
7. A speech decoding apparatus comprising:
-
(a) a decoder for extracting separately character codes and accent identifying codes from a speech information containing Telex signal transmitted and/or recorded in a Telex system; (b) first and second buffer memories for temporally storing the character codes and the accent identifying codes as separately extracted, respectively; (c) a main control circuit for editing said character codes into syllable data; (d) speech element memory means for storing therein a series of the speech elements for each syllable such that the data edited by the main control circuit may be read out in a sequence corresponding to a time series of the speech elements in a natural voice speech; (e) a signal processing circuit responsive to syllable data from said first buffer memory which data constitutes a speech word to be synthesized for providing a read signal for reading a series of corresponding speech element information, a pitch control signal for designating a read-out duration for each speech element information in each of said series of speech element information, and a syllable duration control signal for designating a time duration of each syllable; (f) a speech element information synthesizing circuit responsive to said control signals for synthesizing said series of said speech element information read-out from said speech element memory means into a speech element information train signal corresponding to a word to be synthesized; and (g) circuit means for converting said speech element information train signal into a voice speech signal.
-
Specification