Filipino AI voiceover

The Filipino voices, with samples

Each sample reads the same line so you can compare them fairly.

Two voices, and why

The live catalogue holds no premium characters for fil-PH, so the product falls back to the locale's best tier, Wavenet, choosing C for male reads and A for female. That verification is baked into the casting code, not guessed per request, so generation never fails on Filipino; it just has less texture to offer. Compared with the eight-character languages, delivery is flatter across long passages, though clarity is excellent. The per-scene voice dialog can search the full catalogue, which for fil-PH surfaces Standard-tier voices, older still, plus whatever Google adds over time.

Style requests and where they go

Ask for energetic, warm, instructional or neutral and Filipino returns the same two voices; only the gender preference changes the outcome. The register then lives in the writing, and the director uses it: a promo script gets short exclamations and a sale rhythm, a tutorial gets even, numbered steps, and the tutorial rule still applies in the sense that the script is written instructionally. Taglish is available on request; describe your brand voice in the brief and the mix of Tagalog and English follows it, which is often the most natural register for Metro Manila retail.

Full pipeline, older instrument

Everything around the voice runs at full capability. The Language field set to Filipino makes the director write narration and on-screen copy natively from any brief. Scenes stretch to fit clips, dynamic pace flows overruns across cuts, each scene has its delay in seconds and speed steps from 0.85x to 1.25x, and music ducks itself to 22 percent under speech. A typical clip uses about half a credit, billed by length, and an export bills only its render time once clips are complete. The word-timed captions are actually at their most precise here, since Wavenet supplies timing marks natively.

Getting the best from Wavenet

A flatter voice rewards livelier text. Keep sentences short and rhythmic; the evenness that drones through a fifty word sentence sounds crisp across ten word ones. Spell out numbers whose reading matters, isang daang piso over P100, and respell foreign or English names phonetically in the narration when the read is off, about half a credit per regeneration. Sentence punctuation cuts the caption cues, so punchy sentences also caption cleanly. The em dash is banned in generated narration by prompt rule; commas suit Filipino cadence better anyway.

Captions in this script

Filipino captions are among the simplest cases: spaced Latin words every catalogue font renders natively, so all 21 looks keep their own typography. There is also a timing bonus specific to Wavenet voices: they honour timing marks directly, so the word highlight is measured on the exact clip you hear instead of scaled from a sibling read, and Filipino caption sync is correspondingly tight. Long Taglish sentences past 48 characters split into a second cue, so shortish sentences read best in the pill.

Questions people ask

Related pages