Automated Voice Inflection for Number Strings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current IVR systems produce spoken numbers that sound mechanical due to the lack of proper inflections, making them unacceptable as they fail to mimic human speech patterns, especially when concatenating prerecorded numbers.
Innovation Solution
A system and method that records numbers in groups of specific lengths (2, 3, or 4) to capture natural speech inflections, breaking them down into individual strings with unique inflections for each position, allowing for the retrieval and assembly of properly inflected numbers when spoken.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If prerecorded numbers are concatenated together to form number strings, then the system can speak numbers efficiently, but the spoken numbers sound mechanical and lack proper inflections
Solution Approach 1:
The patent segments number strings into individual digits with separate inflection recordings. Instead of recording complete number strings, the system records each digit (0-9) with different inflections (upward, downward, neutral) and stores them separately. When a number string needs to be spoken, the system retrieves the appropriate inflected digit recordings and assembles them in sequence, creating natural-sounding speech while maintaining efficiency.
2Ease of operation
If a recorder records every possible number string combination, then all number strings can be spoken with proper inflections, but the memory requirement becomes excessively large
Solution Approach 1:
The patent creates a universal set of inflected digit recordings that can be combined to form any number string. By recording each digit (0-9) with multiple inflections (upward, downward, neutral) once and storing these universal building blocks, the system can generate any possible number string combination without needing to store every complete string separately. This dramatically reduces memory requirements while maintaining the ability to speak all number strings naturally.
3Ease of operation
If the system stores inflected digit recordings for all positions and string lengths, then proper inflections can be achieved for any number string, but the device complexity increases
Solution Approach 1:
The patent applies local quality by recording different inflections for digits at different positions within number strings. Instead of using a single inflection for each digit, the system records upward inflections for certain positions, downward inflections for other positions, and neutral inflections for specific contexts. This localized approach to inflection allows the system to achieve accurate, natural-sounding speech for any number string while organizing the data in a manageable, position-based structure.
Data Source
AI summary
The present invention is directed to systems and methods in which a speaker records strings of numbers in different string lengths. Advantage is taken of the fact that speakers typically break numbers into group sizes of two, three, or four. Thus, by way of example, a recorder records two 0's, two 2s, two 3s, etc. Then the recorder records three 1s, three 2s, three 3s, etc., followed by four 1s, four 2s, four 3s, etc. The spoken number values for each string are broken apart and stored as individual numbers corresponding to the string length of the recording. When a number string is to be spoken (for example, the number 782), the system retrieves from the three digit string a first 7, a middle 8, and an end 2. When these retrieved values are communicated to a recipient, proper inflections are achieved for each digit.