1. Choose the Correct Language

This is the most important step. When uploading your video, make sure to select the correct language setting. For example:

For casual Cantonese speech, choose "Cantonese Traditional"
For formal Mandarin narration, choose "Traditional Chinese Written"
For mixed-language videos, select the predominantly used language

Language selection directly affects the AI model's recognition strategy — choosing wrong will significantly reduce accuracy.

2. Ensure Good Audio Quality

AI transcription accuracy largely depends on audio quality:

Reduce background noise — Use a noise-canceling microphone or apply noise reduction in post-production
Avoid music overlapping dialogue — Loud background music severely affects recognition
Maintain consistent volume — Fluctuating volume causes some segments to be missed
Use an external microphone — Much better audio capture than built-in phone mics

3. Leverage AI Proofreading

Hey Subtitle's built-in AI proofreading (powered by AI) automatically:

Corrects common homophone errors
Optimizes sentence segmentation
Fixes grammar issues

Proofreading runs automatically after transcription and typically eliminates most errors.

4. Use Find & Replace

If you notice the AI consistently misspells certain words (like proper nouns), use the editor's Find & Replace feature to fix all occurrences at once instead of editing line by line.

5. Speaking Tips During Recording

If you're still in the recording phase, these habits improve subsequent transcription accuracy:

Speak clearly — You don't need to speak slowly, just avoid mumbling
Minimize filler words — Too many "um", "ah", "like" add noise
Pause between sentences — Brief pauses help AI segment correctly

Summary

Subtitle accuracy is a combination of audio quality, language settings, and AI capability. Get the first two right, and combined with AI proofreading, you'll achieve high-quality subtitle output.