VibeVoice 7B when the advert should be a conversation, not a read
Every other engine reads with one voice. This one takes a script with labelled turns and hands back a single file with two people in it — and we confirmed it counts to four. It also accepts 30,000 characters, three times more than anything else here, so an entire product podcast fits into one job.
Rendered by this model
Turn the sound on and listen
The script the engine was given
Speaker 0: Słuchaj, potrzebuję opisu do nowego kubka jeszcze na dziś. Speaker 1: Zrobione. Trzysta trzydzieści mililitrów, grube ucho, szkliwo do zmywarki. Speaker 0: A pakowanie? Bo klienci pytają. Speaker 1: Tektura, bez folii. I wysyłka tego samego dnia, jeśli zamówią przed czternastą.
Specification
- Built by
- Microsoft
- Speaks Polish
- yes — checked on a Polish sentence
- Ready-made voices
- 8
- Voices in one file
- up to 4
- Steering the delivery
- none — the words are the only lever
- Character limit
- 30,000
- File from the engine
- mp3
- Render time
- about 120s
- Cost
- 1 credit per 1000 characters, rounded up
What it is good for
- Up to four voices in one script
- You label each turn with a speaker number and the engine hands them out between the voices you chose. The question-and-answer shape that sells best in adverts is one job rather than two files stitched in an editor. The four speakers were confirmed by a separate test, because the documentation publishes no limit.
- It speaks Polish despite the English presets
- Each of the eight voices is labelled English or Chinese and that is misleading — it reads Polish correctly, with a Polish accent. It is the most surprising thing in the whole set of ten, which is why we checked it twice.
- Thirty thousand characters per job
- Three times the MiniMax limit and six times ElevenLabs v3. A ten-minute conversation, a description of a whole collection or the script for an about page all fit without being divided.
When to pick something else
It is the slowest of the six Polish engines — a short line took us about two minutes — so it is no use for auditioning takes. There is also no delivery control at all here: no mood field like MiniMax, no sliders like ElevenLabs Multilingual v2 and no tags like ElevenLabs v3. You get whatever the engine does with the sentence, and the only lever is the words themselves. For a single advert read by one voice use MiniMax Speech 2.8 HD; you come here for a conversation.
See every modelHow to brief this model
The things this engine handles differently from the rest. The habits that hold for every model are collected separately, in the help centre.
- Number the turns from zero and keep one turn on one line. A script without line breaks comes back as a monologue even when you have picked two voices.
- Build the mood out of the words, not a tag. This engine reads no emotion tags, so a question asked outright sounds like a question and an exclamation has to actually be written.
- Pick two clearly different voices — male and female, or low and high. Two similar presets in one file sound like one person answering themselves.
- Do not start your search for a delivery here. At two minutes a line, tune the copy on MiniMax Speech 2.8 Turbo and come here for the finished dialogue.
- Ignore the "[EN]" on a voice name. It labels the preset, not the language — a Polish script will be read in Polish.
Questions about this model
Other models
- MiniMax Speech 2.8 HDA Polish voice with an emotion you pick and pauses you place.
- MiniMax Speech 2.8 TurboThe same Polish voice as HD, back sooner. For auditioning takes.
- ElevenLabs Multilingual v2It expands "24 h" into "24 godziny" by itself. For copy with numbers in it.
- ElevenLabs Turbo v2.5The shortest wait of the ten. For short adverts and quick takes.
- ElevenLabs v3You write the mood into the middle of the sentence, in square brackets.
- KokoroEnglish only, and nothing to configure. For adverts abroad.
- Chatterbox HDThe only engine here that will speak in your own voice.
- Orpheus TTSA laugh or a sigh mid-sentence. English, and slowly.
- DiaTwo people out of one text field. No voice to choose.
Try this model on your own product
Sign up without a card and get 40 credits. Switching model is one click, so you can send the same brief through several of them and keep the best frame.