How to Use AI Speech to Text in PowerDirector
Turn what is said in your video into timed text, then fix the few words it mishears.
Updated Sep. 13, 2026 · 4 min read
In a hurry? Use the AI Speech to Text quick guide instead.
PowerDirector's AI Speech to Text listens to your video and writes down what is said, as timed lines on their own track. Our 33-second presenter clip came back as eight lines, each with its own start time. What is left for you is a read-through: fixing the odd misheard word before the text goes on screen or into a file.
- Put your video on the timeline, open Subtitles, and choose AI Speech to Text.
- Pick the Audio source and the Language spoken, then click Create.
- Read the timed lines in the list, or play the preview to check them.
- Click any misheard line to fix it, and press Esc to save.
Steps to Use AI Speech to Text
Step 1: Put the Video on the Timeline and Open Subtitles
AI Speech to Text transcribes what is on the timeline, so add your clip there first. Then click Subtitles along the top.
The room opens on three ways in. Choose AI Speech to Text, the one for a video with speech and no transcript yet.
Step 2: Pick the Audio and Language, Then Create
The dialog asks for two things. Audio source is the track to listen to, which matters when music sits on a track of its own. Language is the language spoken in the clip.
Transcribe selected range only limits the job to a range you have marked. Total duration beside the button shows how much audio will be transcribed, and the badge on Create shows the credits it will use. Check both, then click Create.
Step 3: Read Through the Transcript
When it finishes, the Subtitles panel lists every line with its Start Time, and the same lines sit on a subtitle track above your clip. The preview shows each one as it plays, so you can listen and read at once.
Long sentences are split across lines to fit on screen, which is why line 7 is only three words. That is normal, not an error.
Step 4: Fix the Words It Misheard
Ours got almost everything right. Line 4 is the exception: your own Empower Director is PowerDirector, misheard. Product names and people's names are where this usually happens.
Click the line to edit it. Enter adds a line break and Esc saves, so press Esc when you are done. For a word that is wrong on every line, use Find above the list and Replace All.
The Result
Every word the presenter said is now text: timed, on its own track, and one correction away from right. It plays as subtitles as soon as it exists, and you can also export it as a file. To restyle the lines or move their timing, see how to add subtitles.
Haven't got PowerDirector yet? Drop in an interview, a vlog or a talk, and AI Speech to Text types out what is said as timed subtitles you only need to proofread.
Download PowerDirector Free →FAQ About AI Speech to Text
Yes. The Create button carries a credit badge showing what the run uses, and the dialog has a Get Credits button. Check Total duration first, and use Transcribe selected range only if you need just part of the timeline.
Yes. In the Subtitles room, open the ... menu and choose Export Subtitle. Export Without Style Formatting gives a plain SRT file with the text and its timings.
Press Esc to save a line. Enter puts in a line break instead of confirming, which is the opposite of most text fields.
Set Audio source to the track with the speech on it. The dialog transcribes the track you choose, so the music is left out.