---
title: "The Vocal Speech Understanding System"
authors:
  - "Stephen E. Levinson"
doi: ""
url: "https://ijcai.org/proceedings/1975/78"
pdf: "https://ijcai.org/proceedings/1975/0078.pdf"
venue: "Proceedings of the Fourth International Joint Conference on Artificial Intelligence"
year: 1975
pages: "499-504"
track: "Speech Understanding"
keywords:
---

# The Vocal Speech Understanding System

Stephen E. Levinson

*Proceedings of the Fourth International Joint Conference on Artificial Intelligence*, pages 499-504.

## Abstract

This paper describes the VOCAL (Voice Operated CALculator) speech understanding system. VOCAL is a software package that lets its user program a computer to perform numerical calculations by speaking to it in English-like sentences. To accomplish this, VOCAL uses processes for acoustic, grammatical, and semantic analysis. These individual procedures, which are relatively simple, are embedded in a control structure that uses the information from each component to arrive at a meaningful interpretation of spoken sentences. One unique feature of VOCAL, which is essential to the development of speech understanding systems, is that it is complete and self-contained. Coded in standard FORTRAN, it is compact enough to run on many minicomputers and can be used in a real-time, on-line environment on slightly more powerful machines. Testing has shown that despite a correct word identification rate of less that 60% the VOCAL system usually correctly interprets even very long sentences.