Southeast Asia · Latin · Basa Sunda
Sundanese subtitle generator
Use this Sundanese subtitle generator to transcribe Basa Sunda audio locally in your browser, review the Latin cues and wording, then export SRT, VTT or TXT. The draft needs a fluent human review, and no account, upload or paid API is required.
By Abdul Samad · Last updated
Generate subtitles
Runs Whisper in your browser. No upload, no account, no credit card.
Choose a video or audio file
Drop it here, or click to browse. MP4, MOV, WEBM, MP3, WAV, M4A and similar formats.
Your file never leaves your device
Up to 500 MB. Files over 100 MB can be slow on phones.
Only transcribe media you own or have permission to use. VoiceToTextFree does not download media from other sites.
Auto-detect works for clear speech; choosing the language is faster and safer for accents.
Limits: 500 MB per file; speech must be audible; Whisper can struggle with music, overlapping voices and very noisy recordings. VoiceToTextFree does not upload files or send audio to a paid API.
Key facts
| Item | Detail |
|---|---|
| Language code | su |
| Native name | Basa Sunda |
| Script | Latin |
| Direction | Left to right |
| Region | Southeast Asia |
| Output | SRT, VTT and TXT |
What Sundanese transcription looks like
Sundanese is an Austronesian language of West Java and Banten, written mainly in the Latin alphabet and spoken with distinct speech levels. VoiceToTextFree runs Whisper locally in your browser, drafts Latin-script Sundanese and lets you correct cues before exporting SRT or VTT. The draft may not match the speaker's lemes or loma register, and it does not promise romanized output for mixed Sundanese and Indonesian speech.
Script and punctuation in Sundanese
Modern Sundanese uses the Latin alphabet, including the digraph eu for a central vowel and occasional é. The revived Aksara Sunda script appears on signage and in cultural projects but is not what Whisper returns. Register, rather than spelling, is the main editorial decision.
Common Sundanese transcription pitfalls
Speech levels such as lemes and loma are easy to confuse, especially in pronouns, verbs and greetings that change with the audience. The eu vowel can be misheard as e or u, changing the word, and the model may normalise it to a nearby Latin spelling. Indonesian code-switching is common, and the draft may use Indonesian vocabulary without marking the switch for the reviewer. Banten, Priangan and Cirebon varieties differ in sound and word choice, and the model may drift toward one regional standard. Personal and place names often have variant spellings, and a model cannot check which form the family or community prefers.
Tips for cleaner Sundanese subtitles
- Identify the register early and check pronouns, verbs and greetings against it before you edit individual words.
- Listen for eu and confirm words where e, eu and u would all be plausible, especially in short cues with little context.
- Keep Indonesian insertions explicit rather than silently replacing them, because the audience may expect the original wording.
- Keep a glossary of local names, foods and cultural terms, and apply one spelling throughout the file.
- Ask a native reviewer to check a formal or ceremonial passage where the wrong register would be conspicuous.
- Test a cue with the eu digraph in the target player, since font fallback and encoding can affect how it appears. Compare digits with number words such as hiji and dua, because Indonesian numerals often replace Sundanese ones in speech.
Line length and timing for Sundanese
Sundanese reads well at about 32 to 40 characters per line, with two lines as the practical maximum. The eu digraph and code-switched Indonesian can make a line look longer than its word count suggests. Break at phrase boundaries and keep register markers together. Test a cue with eu and a code-switched word in the player.
Who this Sundanese workflow helps
Suits cultural documentation, local radio, vlogs, school recordings and family videos from West Java. It also helps translators who need a rough Sundanese draft before a native speaker checks register and dialect, and it keeps the recording local while the draft is prepared.
A Sundanese cue with its English meaning
A short line transcribed as native script. Whisper generally returns the language's normal script, not a romanized version.
1
00:00:01,000 --> 00:00:04,000
Isuk-isuk urang papanggih di dieu.
2
00:00:04,200 --> 00:00:08,000
Kamar 402 aya di lantai opat.Frequently asked questions
Does VoiceToTextFree support Sundanese?
Sundanese is present in the multilingual Whisper tokenizer used here, so it can be selected and transcribed in the browser. This page has no reviewed sample, so accuracy is not claimed and a native reviewer should still check register.
Does Whisper output the Aksara Sunda script?
No. Whisper normally returns Latin-script Sundanese. Aksara Sunda is a revived traditional script used in cultural contexts, but the model does not convert to it, and no romanization or script conversion is promised.
What is the difference between lemes and loma?
Lemes is the polite register and loma is the informal one, with different pronouns, verbs and greetings. A transcript can flatten the difference, so decide which register the video uses and check the cues against it.
Why does the eu vowel cause errors?
The eu digraph represents a central vowel that can sound close to e or u to a model trained on other languages. Confirm words where the choice changes meaning, and keep the spelling consistent once you decide.
Can I export Sundanese subtitles as SRT?
Yes. After editing, download SRT, VTT or TXT as UTF-8. Check that é and the eu digraph render correctly, and test a cue with Indonesian code-switching in the target player.
How accurate is Sundanese transcription?
No reviewed Sundanese sample has been recorded, so this page makes no tested accuracy claim. Register, dialect, the eu vowel, audio quality and model size can all affect the result, so treat the draft as a first pass.
Sources and further reading
Abdul Samad — Creator and maintainer. About Abdul Samad.
Last updated · Published
Related guides
- How to improve Whisper accuracy for cleaner subtitlesHow to improve Whisper accuracy with better audio capture, explicit language selection, measured model sizes, overlap handling and a practical human review workflow.
- Right to left subtitles and non-Latin scripts: a practical guideRight to left subtitles need UTF-8, bidirectional text, correct punctuation, suitable fonts and player testing. A practical guide to RTL SRT and WebVTT.
- Subtitle best practices: a practical guide to cleaner captionsSubtitle best practices for real playback: clause-level segmentation, two-line limits, reading speed, punctuation, speaker IDs, SDH and a clean review pass.
Other languages
Languages in the picker: 99. The list is checked against the Whisper tokenizer by a unit test.