Transcribe Filipino audio to text, free

What makes Filipino hard to transcribe

Everyday Filipino speech is Taglish — Tagalog grammar carrying English vocabulary, switching several times within a sentence. Unlike code-mixing in other languages, the switch is so frequent that treating the recording as either pure Tagalog or pure English produces a worse result than accepting the mix.

What your Filipino transcript looks like

Tagalog words and English words both appear in Latin script, so the transcript reads the way Filipinos actually write online.

[00:00 → 00:04] Magandang umaga sa lahat, simulan na natin ang meeting.

[00:04 → 00:08] Una, pag-usapan natin ang results ng last month.

An illustration of how Filipino transcripts are formatted — script, punctuation and timestamps. Not the output of any particular recording.

How accurate to expect

Taglish is the normal input and transcribes acceptably. Pure formal Tagalog, which almost nobody speaks casually, is actually rarer in the training data than the mixed register.

What usually needs fixing

1. 1 Deciding whether a borrowed word should be spelled the Filipino way or kept in English.
2. 2 Spanish-derived vocabulary, which has both Filipino and Spanish spellings in common use.
3. 3 Regional language passages in Cebuano or Ilocano, which are not Filipino and will not transcribe correctly.

Getting better results

• → Do not try to force the language to English even when a recording sounds mostly English — the Tagalog grammar words will suffer.
• → Regional languages like Cebuano and Ilocano are separate languages, not Filipino accents.

Who transcribes Filipino