CTC (Alignment)
CTC helps match speech sounds to text even when people speak at different speeds.
What it does
CTC helps match speech sounds to text even when people speak at different speeds.
Alignment: Matching timing of sound to text • CTC: Connectionist Temporal Class.
How it works
- Look at sound pieces over time.
- Allow repeats and blank spaces.
- Collapse them into the final text.
Uses & Applications
- A caption app can handle someone saying “hellooo” slowly.
- A speech-to-text system can match fast or slow talking.
- A voice assistant can line up sounds with words.
In short
It aligns sound with text when the timing is not exact.
Try it yourself
Turn repeated sound guesses into a word
A speech system may repeat a letter over several moments. This lesson cleans up the repeats and blank spaces.
Your turn
Change one setting at a time. Reset example puts everything back.
Collapse repeated labels and blanks
Step 1Why this happens