Project Gutenberg's Emotionless AI Audiobooks
Project Gutenberg used AI technology to produce 5,000 audiobooks, showing how AI is increasingly being integrated into book publishing. However, the AI-produced audiobooks lack emotion and variation compared to human-narrated audiobooks.
Summary
- Project Gutenberg collaborated with Microsoft and MIT to produce 5,000 AI audiobooks using neural text-to-speech and other AI technologies to automate audiobook production.
- The typical audiobook production process is lengthy and labor-intensive. AI automation greatly sped up the process for Project Gutenberg.
- The AI cloned and recreated voices from samples to narrate the books. It added emotions to the words.
- Project Gutenberg produced more audiobooks in a short time than a major audiobook producer does per year.
- The AI voices sound human-like but are flat, emotionless, and lack variation. There is no differentiation based on genre or characterization.
- The lack of good narration and dramatic elements makes the listening experience unstimulating compared to human-narrated audiobooks.
- The catalog includes many classics by white authors. More diversity is needed.
- AI audiobooks can benefit those unable to afford regular audiobooks, which are expensive.
- AI narration could help disabled people access more books if developed considerately.
- AI narrators may threaten jobs, but human narrators likely still have some time before AI matches their quality.
- AI voice technology will continue improving over time. Safeguards may be needed for commercial use.