Speech-controlled electronic apparatus

• US 4,573,187 A
• Filed: 05/17/1982
• Issued: 02/25/1986
• Est. Priority Date: 07/24/1981
• Status: Expired due to Fees

• Current Assignee: Asulab SA (The Swatch Group AG)
• Original Assignee: Asulab SA (The Swatch Group AG)
• Inventors: Monbaron, Jean-Jacques, Michel, Jean-Georges, Bui, Ngoc C.
• Primary Examiner(s): Kemeny, E. S. Matt
• Application Number:
  US06/379,265
• Time in Patent Office: 1,380 Days
• Field of Search: 381/41-43, 364/513, 364/513.5
• US Class Current: 704/275
• CPC Class Codes: G04G 21/06 using voice
• CPC Class Codes: G10L 15/00 Speech recognition G10L17/00 takes precedence

Abstract

In a speech recognition and control apparatus, the speech signal is filtered into sub-bands, each sub-band signal sampled and binary quantized to form a digital string which is simplified (data reduced or compressed) by further encoding based on detecting in the string an isolated sample having one binary value surrounded by two samples having the other binary value.

Claims

1. A speech-controlled apparatus for performing a plurality of functions comprising:

electro-acoustic means for converting a word pronounced by the user of the apparatus, said word belonging to a predetermined vocabulary, into an analog signal;

means for coding said signal including;

a plurality of band-pass filter means for filtering said analog signal, each of said band-pass filter means having a distinct band-pass to provide a plurality of filtered analog signals,

means for sampling each of said filtered analog signals,

means for setting a threshold value, and

means for comparing each sample of each filtered signal to said threshold and for allocating to each sample a first digital value if said sample has an amplitude which is greater than said threshold, and a second digital value in the opposite case, thereby forming from each analog filtered signal, strings of digital values, said strings forming together said coded digital signal;

said apparatus further comprising;

first processing means for sequentially detecting in each of said strings an isolated sample having said second digital value surrounded by two samples having said first digital value and for allocating to said isolated sample said first digital value, and means for then sequentially detecting in each of said strings an isolated sample having said first digital value and for allocating said second digital value to said isolated sample, whereby simplified strings of digital signals are formed, said simplified strings of digital signals forming together a coded digital signal representative of said pronounced word;

means for memorizing as refernces in coded form all of the words of said predetermined vocabulary, each reference corresponding to an instruction for performing at least certain functions of the apparatus;

means comparing said coded digital signal to at least a part of said references;

means for selecting the compared reference which is closest to said coded digital signal;

means for informing the user on the selected reference in response to said selected reference;

means for transforming said selected reference into a control signal; and

actuating means responsive to said control signal for performing said functions.

2. The apparatus of claim 1 further comprising second processing means comprising means for sequentially comparing the digital value of each of the plurality of simplified strings corresponding to one sampling moment to the value of the simplified strings corresponding to a preceding sampling moment, means for producing a transition signal whenever the comparison means detect a difference, means for counting the number of sampling moments after the production of a transition signal without a fresh transition signal appearing, and means for suppressing the samples of the simplified strings after said number has reached a predetermined value and until a fresh transition signal appears, whereby said simplified strings are converted respectively to normalized strings forming together said coded digital signal.

3. The apparatus of claim 2, further comprising means for memorizing the conserved samples of the plurality of normalized strings corresponding to the same sampling moment if said samples are overall different from those of the preceding sampling moment and for allocating to each memorized group of samples a binary value representing the position of the memorized group of samples with respect to the chronological succession of the groups of conserved samples.

4. The apparatus of claim 1, wherein the performance of each of said functions requires the introduction by uttering corresponding words of said vocabulary of a cycle of instructions, comprising at least an instruction for selecting one of said functions and at least two instructions, representative of numerals to be used in said apparatus, further wherein said reference memory means comprise means for memorizing the group of references associated with the function selection instructions and with the instructions representative of said numerals, and wherein said means for producing control signals comprise:

means for transforming each selected reference into a digital signal in binary form; and

means for associating with each digital signal in binary form a signal representative of the time position of the instruction corresponding to said signal in the cycle associated with the performance of said function.

5. The apparatus of claim 4, wherein said means for associating a time position signal with each signal in binary form comprise:

means for producing a starting pulse synchronously with each selected reference, said starting pulse being also representative of the fact that a word has been pronounced;

a circuit for receiving each of said starting pulses in order to supply a sequencing pulse if no starting pulse has been applied to said circuit within a given time interval, said sequencing pulse being representative of the position of each of said instructions in the succession of instructions required for performing said function; and

means for counting said starting pulses and for supplying said time-position representing signal, the digital value of which is equal to the content of the counting means and for resetting said counter to zero when the number of starting pulses applied thereto is equal to the number of instructions of a cycle.

6. The apparatus of claim 5, wherein said vocabulary comprises in addition to said function selection instructions and said instructions representative of numerals to be used in said apparatus, instructions for starting and stopping said selected function, said instructions in the same cycle being introduced in the order specified so that each instruction corresponds to a given value of the time-position representing signal in an increasing order, said means for producing the control signals further comprising means for memorizing the value of the digital signal in binary form when the time-position representing signal is of the value corresponding to the introduction of the function selection instructions, thereby generating a mode control signal, and means for supplying an information control signal when the time-position representing signal is of a different value.

7. The apparatus of claim 1, wherein said informing means comprise display means for displaying at least a part of said words.

8. The apparatus of claim 1, wherein said informing means comprise sound generating means representative of at least a part of said words.