VOICE SYNTHESIZER WITH DIGITALLY STORED DATA WHICH HAS A NON-LINEAR RELATIONSHIP TO THE ORIGINAL INPUT DATA

• US 3,803,358 A
• Filed: 11/24/1972
• Issued: 04/09/1974
• Est. Priority Date: 11/24/1972
• Status: Expired due to Term

• Current Assignee: Sheldon Apsell, Vincent Schirf
• Original Assignee: Sheldon Apsell, Vincent Schirf
• Inventors: Schirf, Vincent, Apsell, Sheldon
• Primary Examiner(s): Claffy, Kathleen H.
• Assistant Examiner(s): Leaheey, Jon Bradford
• Application Number:
  US05/309,088
• Time in Patent Office: 501 Days
• Field of Search: 179/1SA,1SB,15.55T 340/148,152
• US Class Current: 704/267
• CPC Class Codes: G10L 13/02 Methods for producing synthetic speech; Speech synthesisers

Abstract

An electronic speaking machine has its vocabulary stored in a solid state memory so that the device, with the possible exception of the sound generator, employs no moving parts. The machine is capable of reproducing any spoken word by storing a digital representation of that word in its vocabulary. To reduce storage space, data compression is employed to reduce the data obtained from sampling an audio signal of the spoken word. Because only fixed words are stored, the data compression technique employed can be optimized for each stored word. A particular word is selected by applying the proper '"'"''"'"''"'"''"'"'select code'"'"''"'"''"'"''"'"' to the input of the apparatus. A '"'"''"'"''"'"''"'"'start of word'"'"''"'"''"'"''"'"' signal then causes a clock to sequence a counter through the addresses in the memory where the digital data representing the word is stored. Inasmuch as the stored digital data has a non-linear relationship to the original data, the non-linear data read out of the memory is transformed by a non-linear mapper to digital data having a linear relationship to the original data. A digital to analog converter transforms the linear digital values into an audio signal that is then filtered to obtain a reconstruction of the original audio signal of the spoken word. The reconstructed audio signal can then be used as the input to a conventional amplifier and speaker system.

Claims

1. An automated voice response system comprising a memory device having a vocabulary of spoken words recorded thereon, each spoken word being recorded in the memory device as a sequence of digitally coded numbers non-linearly related to selected amplitude levels of an audio waveform derived from the spoken word, the number of selected amplitude levels being such as to provide a substantial reduction in digitized data obtained from sampling the amplitude of the audio waveform, the digitally coded numbers representing at least some of the words in the vocabulary being differently related to the selected amplitude levels of their audio waveforms whereby a different non-linear relationship exists for those words, a decoder for enabling any word in the vocabulary to be read out of the memory device in response to a word select command, means for sequentially reading out of the memory device the digitally coded numbers representing the selected word, a plurality of gates controlled by the decoder, a plurality of non-linear mappers, each non-linear mapper having its input coupled through a different one of the gates to the output of the memory device whereby when the gate is enabled the non-linear mapper receives digitally coded electrical signals from the memory device, each non-linear mapper converting received digitally coded electrical signals to coded electrical output signals whose numerical values are linearly related to the selected amplitude levels of the audio waveform of at least one of the vocabulary words, and a digital to analog converter responsive to the outputs of the non-linear mappers for converting the linearly related coded electrical signals emitted by those non-linear mappers into equivalent analog signals.

2. The automated voice response system according to claim 1, wherein the aforesaid vocabulary words having different non-linear relationships are associated with different ones of the non-linear mappers, and wherein the decoder, upon enabling a selected word to be read out of the memory device, also enables one of said plurality of gates whereby the output of the memory is fed into the non-linear mapper associated with the selected word.

3. An automated voice response system comprising a memory device having a vocabulary of spoken words recorded thereon, each spoken word being recorded in the memory device in the form of a sequence of digitally coded numbers representing non-linearly related selected amplitude levels of the audio waveform of the spoken word, the number of such selected amplitude levels providing a substantial reduction in data obtained from sampling the amplitude of the audio waveform, a decoder for enabling a word to be read out of the memory device in response to a word select command, means for sequentially reading out of the memory device the digital codes representing the selected word, said means including a rate selector for setting the rate at which read out is effected, a non-linear mapper having its input coupled to the output of the memory device and receiving therefrom digital coded electrical signals, the non-linear mapper providing a mapping output which converts the received digital coded electrical signals to coded electrical signals whose numerical values are linearly related to said selected amplitude levels of the audio waveform, a digital to analog converter having the output of the non-linear mapper coupled to its input, a filter coupled to the output of the digital to analog converter for smoothing the output of the converter, the filter being of the type having a variable pass band, and rate detector means coupled to the memory device, the rate detector means being adapted to ascertain the appropriate rate for reading information out of the memory device, the output of the rate detector controlling the rate selector and the pass band of the filter.

4. In an automated voice response system of the type employing a memory device having recorded in it a vocabulary of spoken words, each word being recorded as a sequence of encoded numbers representing the amplitude at sampled points of an audio waveform derived from the spoken word, a decoder for enabling a word to be read out of the memory device in response to a word select command, means for sequentially reading out of the memory device in the form of digital electrical signals the encoded numbers representing the selected word, a digital to analog converter for converting digitally encoded electrical input signals to output signals which are analogs of the numerical values of the encoded signals, the improvement for compressing the sample data to enable a word to be intelligibly reproduced with a substantial reduction in information recorded in the memory device, wherein in the recorded sequence of encoded numbers representing a word, the encoded numbers are non-linearly related to selected amplitude levels of the sampled audio waveform, and wherein the automated voice response system further includes a non-linear mapper having its input coupled to the output of the memory device and receiving therefrom coded electrical signals representing the selected word, the non-linear mapper having its output coupled to the digital to analog converter, and the non-linear mapper responding to the electrical signals from the memory device by emitting coded electrical signals whose numerical values are linearly related to the aforesaid selected amplitude levels of the sampled audio waveform.

5. In the automated voice response system according to claim 4, the further improvement wherein the non-linear mapper is arranged to provide different mapping outputs, and the decoder includes means for selecting the mapping output provided by the non-linear mapper.