A support team needs written transcripts of recorded conversations. Which capability directly fits?

1. Speech recognition
2. Sentiment analysis alone
3. Image generation
4. Speech synthesis