Sentioscope

Speech and voice controls, as each vendor documents them

The speech table, and what each column asks of a vendor

One table holds the five fields these notes keep for every model, with what a voice is built from and how it attaches to a character sharing a column. The table is the fastest way in; the model notes behind it carry the wording each cell was read from. As of 2026-09-12.

The questions the table puts to every model, in orderCan it speak, can it speak this language, can it speak as a chosen character, does that voice stay with the character, and what makes the mouth follow the line. A production has to answer all five before dialogue is written, and each is a separate field because a model can pass one and fail the next.Sound atallAudioproducedwith thepicture, orsilentfootagewaiting fora dub.If yesThislanguageWhichlanguagesthe vendornames forspokendialogue,not formenus.If yesA chosenvoiceWhether areferencesample cansteer whothe speechsoundslike.If yesKept on thecharacterWhetherthat voicestays withonecharacteracrosscalls.If yesThe mouthWhat thevendor saysdriveslip-sync,if it saysanything.
Fig. 1 Nearly every entry answers the first question and the later ones thin out, which is what makes their gaps worth printing.

Five fields, chosen because they are the five a production has to settle before writing dialogue. Whether audio is generated with the picture or attached afterwards. Whether a reference voice can be supplied. Whether a voice stays attached to one character. Which languages the vendor names for speech. What, if anything, is documented about lip-sync.

The fields were not chosen to flatter any model. Only the audio column is filled for nearly every entry; the others thin out, and the languages column is empty for most. The gaps are the most useful thing the table shows: speech is the least documented part of generative video.

Reading the table from left to right answers a practical question in order — can this model speak, can it speak this language, can it speak as a particular character, and what makes the mouth follow the line.

1The table

  • Speech — Sixteen models on five speech fields: where audio comes from, what a voice is built from and how it attaches to a character, languages, and lip-sync.

A column is added when enough vendors document something to compare; one nobody addresses stays a question in the notes.

Other notes: Models, Fields, Routes, Side by side, Questions, Terms, Learn, Data. What counts as a documented control is set out on the reading page.

2How one tool handles speech

For the speech side of “Speech controls”, SceneMixer's guides describe how lines are spoken:

SceneMixer writes dialogue into the script analysis, assigns each line to a shot with a speaker and a short delivery note, and the video model performs it — picture and audio generated together, so the mouth forms that sentence in the language you chose.

SceneMixer, AI Dialogue Generator (read 2026-10-13)

A producer working in German can make a series whose cast speaks Indonesian: the interface and the working documents follow the interface language, the spoken dialogue follows the project setting.

SceneMixer, AI Short Drama in 15 Languages (2026): Interface, Scripts and the Dialogue Your Cast Actually Speaks — “Can the interface be in one language and the dialogue in another?” (read 2026-10-13)