Speech-to-Text APIs: Their Advantages and Limitations
summary
The talk examines speech-to-text APIs and where they succeed or fail for tasks like dictation, message writing, podcast transcription, video subtitling and live captioning for meetings. It analyzes which technologies and situations make automatic transcription effective and which do not, and explains why people experience unequal results with speech-to-text systems.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 4:36API technical demo
- 11:54Bias and atypical voices
- 15:48French language challenges
- 23:42Censorship and profanity
- 26:12Punctuation importance
- 28:42Speaker detection
- 30:35Voice recognition improvements
- 35:33Digital accessibility and RGAA
- 36:49France.tv lab experience
- 39:14Limitations of US APIs
- 41:45Subtitling tools and resources
- 42:33Q&A
related talks
56mSpeech Recognition: How to Build Speech-to-Text Yourself
26mMake Python Talk, Make Python Listen
48mYour Apps Can Talk! Introducing Cloud Text-to-Speech, Powered by WaveNet Technology
35mHaving a Voice on Android: How hard can it be?
08mInterview of Emmanuelle Aboaf
25mBeyond Transcription: Building Voice AI That Understands Conversations
58mAdding Speech to your Web App_UIArchitecture
46mArtificial intelligence to the rescue of accessibility