Capture-Time Feedback for Recording Scripted Narration
summary
A tool called Narration Coach records line-by-line, runs ASR plus global and phoneme alignment to aligns takes to the script, and gives text and audio feedback for emphasis, flow, variety, diction. It detects pitch accents with AutoB, modifies pitch/volume/duration with PSOLA (PASOLA), flags awkward pauses and tempo deviations, and estimates diction via video-based mouth-motion tracking. A five-user study and Mechanical Turk ratings show users prefer Coach recordings; emphasis resynthesis often sounds implausible and needs better speech manipulation.
AI-written summary · about
chapters (16)
- 0:00Intro
- 0:15Audio story creation pipeline
- 0:38Good narration
- 1:41What makes narration good?
- 1:58Vocal attributes Emphasis
- 3:39Goal
- 4:13System outline
- 4:45Import script & annotate emphasis
- 5:18Annotate script emphasis (automatic)
- 5:52Record speech
- 9:08Emphasis resynthesis
- 10:27Flow detection & resynthesis
- 11:27Diction feedback
- 12:12Evaluation
- 14:00Feedback & limitations
- 14:41Thank you
related talks
18mFacilitating Development of Pragmatic Competence through a Voice-driven Video Learning Interface
24mFail fast, fail cheap, making voice apps the guerrilla way
44mSpoken vs written dialogue: finding the best voice for your characters
19mInterrupting is the future
50mDesigning Experiences For Voice Interactions
32mPrototyping Voice Experiences: Design Sprints for the Assistant
50mI Think I Need a Writer: Tools for Troubleshooting Narrative
38mAI-Powered Conversational Interfaces