This book is licensed for the personal use of the buyer. Please do not share or resell the file. Short quotations with attribution are welcome.
The information in this book is provided for education. It is not legal, financial or other professional advice. Product and company names are used only to describe them and belong to their owners.
This is a free sample: the introduction and first chapter.
IntroductionIntroduction
Voice is the most natural interface humans have. We speak before we read, and we recognise a friend's voice, a crying baby or a failing engine instantly. Computers have long struggled with sound, but in recent years speech recognition became accurate enough for everyday dictation and captioning, synthetic voices became hard to tell from real ones, and audio models learned to recognise everything from bird songs to machine faults.
This book explains how machines hear and speak. It starts from first principles (what sound is, how it is recorded and how it is turned into pictures called spectrograms) and builds up to speech recognition, speech synthesis, speaker recognition, sound classification, voice assistants and translation. Throughout, it pays attention to what goes wrong: accents and noise, latency, privacy, and the serious ethical questions raised by voice cloning.
No signal-processing background is assumed.
How the book is organised
Chapters 1 to 8: sound, sampling, file formats, loudness, spectrograms and audio features.
Chapters 9 to 17: speech recognition: how it works, data, evaluation, fairness, noise, formatting and real-time use.
Chapters 18 to 23: speech synthesis, expressiveness, voice cloning, synthetic speech detection and speaker recognition.
Chapters 24 to 33: audio classification, sound events, music, translation, voice assistants, design, accessibility, privacy, on-device audio and embeddings.
Chapters 34 to 36: augmentation, deployment and monitoring.
Chapters 37 to 41: contact centres, medicine and law, languages with little data, language models with speech, and emotion recognition.
Chapters 42 to 44: case studies and a project checklist.
Chapter 1Why Audio AI Matters
Audio AI is already everywhere:
Dictation and captions on phones, video platforms and video calls.
Voice assistants in homes, cars and phones.
Transcription of meetings, interviews, lectures and medical consultations.
Accessibility: live captions for deaf and hard-of-hearing people, screen readers and synthetic voices for people who cannot speak.
Industry: monitoring machines, detecting leaks, counting wildlife and analysing calls in contact centres.
Media: dubbing, audiobooks and music tools.
Each application has different demands for accuracy, speed, robustness and privacy.
You have reached the end of the free sample. The full book continues with 43 more chapters, plus the appendices.