Khmer results: an honest view
Khmer has little training data in the speech model, so expect a rough draft with many corrections. It is most useful on slow, clear speech from one speaker. Fast conversation, karaoke-style music, noisy markets and phone calls give much weaker text. Names and Pali or Sanskrit-derived religious terms are frequent errors.
For long interviews, a timed draft to correct is still usually faster than typing Khmer by ear.
Upload the recording as it is: MP3, M4A, WAV, OGG, OPUS, FLAC, a voice message or a phone recording. You can also record from your microphone in the upload box. Nothing needs converting first.
Khmer script in the transcript
The transcript is written in Khmer script. Things to expect:
- Khmer does not separate words with spaces, so lines may be long runs of text, or spaces may appear where the model thought a phrase ended.
- Subscript consonants (coeng) and vowel signs are the characters most likely to be wrong.
- Khmer punctuation such as ។ may be missing or replaced by Latin punctuation.
- English and French words may appear in Latin letters or in Khmer script.
Accents, registers and neighbouring languages
The model is closest to the Phnom Penh standard heard on radio and television. Battambang and other provincial accents, and the Khmer of northeast Thailand (Khmer Surin) and the Mekong Delta in Vietnam (Khmer Krom), differ in pronunciation and vocabulary and need more correction. Royal and monastic vocabulary is used when speaking to or about monks and royalty; dharma talks have many such words.
Choose Khmer for mostly Khmer recordings. Auto-detect may confuse Khmer with Thai or Lao on short clips, so set it yourself.
The transcript opens next to a player that highlights the current line. Click a line to hear it again, correct the word in place, and your changes save automatically.
Who transcribes Khmer
Common files are voice messages from family in Cambodia, oral histories and testimonies, interviews with Cambodian communities in Long Beach, Lowell, Melbourne and France, dharma talks at temples, and NGO or research interviews. Some families record elders telling stories from before and during the Khmer Rouge years so that grandchildren abroad can read them.
Export TXT, DOCX, PDF, CSV, SRT or VTT, with timestamps and speaker names switched on or off per export. Speaker labels are available on paid plans or with a Minute Pack.
For video files, see the Khmer video to text page.