Orpheus TTS when the advert should sound like a phone recording, not a spot
The adverts that work on Instagram today sound like a voice note: someone laughs halfway through a sentence, sighs, or trails off. This engine reads tags written into the copy and does exactly that. It does not speak Polish and it is the slowest of the ten, so it is worth knowing what you are reaching for it for.
Rendered by this model
Turn the sound on and listen
The script the engine was given
Our new collection is finally in. <chuckle> I know, we said two weeks ago. Go and look — the linen shirt is back, and this time in every size.
Specification
- Built by
- Canopy Labs
- Speaks Polish
- no — reads Polish text phonetically
- Ready-made voices
- 8
- Steering the delivery
- tags written into the script
- File from the engine
- wav — re-encoded to mp3 for this page
- Render time
- about 170s
- Cost
- 1 credit per 1000 characters, rounded up
What it is good for
- A laugh, a sigh and a cough as tags
- You place them at a particular point in a sentence rather than setting them across the read. It is the only English-only engine here that can do anything like that mid-utterance.
- It sounds like a person, not a radio spot
- Eight voices with a distinctly less announcer-like timbre than Kokoro. For the format where an advert pretends to be a recommendation from a friend, that imperfection is the point.
- A field for syllable repetition
- The engine can stick on a syllable, so it has a parameter that damps it. Ours is set to the value where the stutters go away without the read turning monotone.
When to pick something else
The slowest of the ten by a wide margin — a single sentence on a cold engine took us nearly three minutes. It is no use at all for auditioning: write the copy elsewhere and come here for one recording. It will not read Polish either; our Polish sentence came out such that speech recognition guessed Portuguese. If you want a clean English read with no effects, Kokoro does it in seconds; if the advert has to be in Polish, use MiniMax Speech 2.8 HD.
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.
- Put the tag where a person would actually laugh — on an admission, not on the offer. A laugh on the price sounds like laughing at the customer.
- One tag per recording. Three laughs in three sentences turns the advert into a sketch and the trick becomes audible.
- Plan for the wait. This is the slowest of the ten, so finish the copy first — on Kokoro if you want to hear it in English straight away.
- Check after a render that the tag was not read aloud. Ours performed <chuckle>, but with a new tag that is the one thing that can ruin an entire recording.
- Do not type Polish here. This is an English engine and a Polish sentence comes out of it sounding like Portuguese.
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.
- VibeVoice 7BTwo people talking in one file. In Polish.
- KokoroEnglish only, and nothing to configure. For adverts abroad.
- Chatterbox HDThe only engine here that will speak in your own voice.
- 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.