Transcribe Thai Audio
Thai is written without spaces between words and carries meaning through tone, so working out where one word ends and the next begins is the whole challenge of transcribing it by hand. ScribeToAny turns Thai speech into properly segmented, readable Thai script for interviews, videos, lectures and meetings, then exports it in eight formats with timestamps.
- AI transcription
- 98 Languages
- Encrypted; we never train AI on your files
Start with your recording
Drop an audio or video file here, or click to choose.
What you get: a real transcript
Unedited output from a 64-second interview with two speakers (synthetic voices), transcribed with speaker detection on. Only the first lines are shown.
SPEAKER_00 (0:00) Welcome back to Small Shop Stories. SPEAKER_00 (0:02) Today I'm talking to Maya, who runs a tiny bakery on the SPEAKER_00 (0:05) corner of Mill Street. SPEAKER_01 (0:07) Thanks for having me. SPEAKER_01 (0:08) It's really small, just two ovens and a counter. SPEAKER_00 (0:12) So how does a normal day start for you?
1 00:00:00,000 --> 00:00:01,680 [SPEAKER_00] Welcome back to Small Shop Stories. 2 00:00:02,160 --> 00:00:05,320 [SPEAKER_00] Today I'm talking to Maya, who runs a tiny bakery on the 3 00:00:05,320 --> 00:00:06,320 [SPEAKER_00] corner of Mill Street.
Checked against the script, 4 of 189 words differed — a 2.1% word error rate, counting "80" written for "eighty" as an error — and all 14 speaker turns were labelled correctly. This was clean speech; noisy recordings and people talking over each other score lower. See the full transcript and how it was scored
How it works
- 1
Upload
Drop in an audio or video file, or paste a link. All common formats are accepted.
- 2
Transcribe
AI transcribes with timestamps and optional speaker labels, usually in minutes.
- 3
Export
Edit segments online, then export TXT, SRT, VTT, TSV, CSV, JSON, PDF or DOCX.
Why ScribeToAny
- Tonal Thai: Output is Thai script with tone marks; noisy audio is where a wrong word slips in.
- Word Segmentation: The model segments unspaced Thai into words as it transcribes; skim word boundaries in ambiguous passages.
- Polite Particles & Registers: Polite particles (ครับ/ค่ะ) and formal vocabulary are kept as spoken.
- Thai Subtitles: Time-coded SRT/VTT; cues use the general 84-character cap, so split long Thai cues in the editor.
Frequently asked questions
How does it split Thai into words without spaces?
The model segments the speech into words as it transcribes, producing Thai script that reads the way native text does. Word boundaries are the thing to skim in the editor if a passage is ambiguous — a click replays the audio so you can check.
Is the output real Thai script?
Yes — the transcript is in Thai script (ภาษาไทย), not a romanised approximation, so it pastes into documents and search works as expected. If you need it in another language, you can translate it into 134+ languages for free, on every plan.
Can it subtitle Thai videos?
Yes. Transcripts are timestamped, so export SRT or VTT and put them straight on the video. The Thai transcript can also be translated into 134+ languages for free, on every plan. The translation keeps the original timings, so a translated SRT lines up with the video too.
Thai isn't in the language menu — how do I make sure it's detected?
Leave Audio language on Auto-detect: Thai is recognised automatically. Detection listens to roughly the first 30 seconds, so if your recording opens with an English intro, music or a long silence, trim that part first — otherwise the file may be treated as another language.
How are Thai subtitles split into lines?
Thai uses the general subtitle limits — about 7 seconds and 84 characters per cue, with breaks at punctuation where there is any. Because Thai runs without spaces and has little punctuation, a cue can come out long for comfortable reading; split long cues in the editor, and switch on Reading-speed & line checks to spot the ones that run too fast.