Speech Compression and Synthesis
Abstract
This document reports progress in the development of a phonetic speech synthesis algorithm, implementation and development of a real-time LPC vocoder, and development of a new high-frequency regeneration method for the excitation signal of voice-excited coders. The speech synthesis system now produces speech using an input string consisting only of triplets of phoneme identity, pitch, and duration, as required for the goal of very low transmission data rate. In real-time vocoder development, the implementation of an effective vocoder testing and debugging program and the implementation of the first two planned vocoder algorithm modifications are reported. The high-frequency regeneration method for voice-excited coders produces quite good voice quality at the cost of little or no extra computation. Keywords include: Speech synthesis, phonetic synthesis, diphone, LPC synthesis, vocoder, speech compression, linear prediction, voice-excited coder, and high-frequency regeneration.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1978
- Accession Number
- ADA155416
Entities
People
- J. Klovstad
- J. Makhoul
- Jean Wolf
- L. Cosell
- Robert E. Schwartz
Organizations
- BBN Technologies