Speech Coding and Synthesis
The fields of speech coding and synthesis have developed rapidly over the last decade. Text-to-text speech systems now produce reasonable quality speech, and currently available speech coders can transmit good quality speech at below 10kb/s. This, in combination with the ever-increasing speed of microprocessors and signal processing hardware, has resulted in a large number of practical applications. These applications in turn have stimulated research, and the number of papers published on speech coding and synthesis have proliferated rapidly. Reflecting periodically on such developments have inspired the publication of this book. Topics such as the effect of cross channel errors on coded speech and the determination of a proper pitch contour for synthesized speech are included.
Both readers unfamiliar with the fields of speech coding and speech synthesis as well as those already working within the areas, will find the book of interest.
Why Read This Book
You should read this book if you want a compact, research‑level survey of mid‑1990s speech coding and synthesis techniques, including low‑bit‑rate coders and synthesis prosody issues. It collects contributions from leading researchers, so you will get practical perspectives on tradeoffs, evaluation methods, and implementation issues that influenced modern codecs.
Who Will Benefit
Researchers and practicing engineers in speech and audio who need a historical and practical survey of low‑bit‑rate speech coding and synthesis methods and their evaluation.
Level: Advanced — Prerequisites: Solid DSP fundamentals (sampling, filtering, z‑transform), basic linear prediction (LPC), probability/statistics, and familiarity with digital filter and transform concepts.
Key Takeaways
- Describe and analyze LPC‑based and CELP‑style parametric speech coders and their tradeoffs
- Evaluate speech quality using objective and subjective measures and interpret results
- Apply pitch and prosody modeling techniques for improved synthesis naturalness
- Assess the impact of channel errors and design basic error‑concealment strategies
- Compare waveform, transform and parametric coding approaches for low bit‑rate targets
Topics Covered
- 1. Introduction: history and trends in speech coding and synthesis
- 2. Speech production models and signal representation
- 3. Parametric coding: LPC and CELP fundamentals
- 4. Waveform and transform coding approaches
- 5. Low bit‑rate coding techniques and algorithms
- 6. Pitch analysis, prosody modeling and synthesis
- 7. Text‑to‑speech synthesis methods and evaluation
- 8. Perceptual models and speech quality assessment
- 9. Effects of channel errors and packet loss on coded speech
- 10. Error concealment and robust coding strategies
- 11. Implementation issues and computational tradeoffs
- 12. Conclusions and directions for future research
Languages, Platforms & Tools
How It Compares
This book is an edited mid‑1990s survey; compared to Quatieri's Discrete‑Time Speech Signal Processing (2001), Kleijn's volume is more of a collection of applied research papers focused on coding and synthesis rather than a single, unified tutorial. Kondoz's Speech Coding (textbook style) gives a more systematic, modern textbook treatment, while Kleijn provides breadth and contemporary research viewpoints from that era.
