ElevenLabs v3 is now available in HyperTTS 2.10. This is a new model from ElevenLabs which allows controlling the emotion of the generated speech using special keywords: https://elevenlabs.io/v3.
There is also a change with the FPT.AI Vietnamese TTS service (see below).
To use the new model, you have to pick one of the v3 (alpha) voices in the ElevenLabs service. Different models appear as a variant of the same voice.
.png)
The difference between models is explained here: https://elevenlabs.io/docs/models test. Some models like Flash 2.5 and Turbo 2.5 are optimized for latency (they will generate faster). If you are batch-generating, this generally won't be a concern for you and you will instead want to pick the higher quality models such as Multilingual v2 and Eleven v3. The new v3 model has two significant improvements.
From the ElevenLabs v3 model page:
The Eleven v3 model supports 70+ languages, including:
Afrikaans (afr), Arabic (ara), Armenian (hye), Assamese (asm), Azerbaijani (aze), Belarusian (bel), Bengali (ben), Bosnian (bos), Bulgarian (bul), Catalan (cat), Cebuano (ceb), Chichewa (nya), Croatian (hrv), Czech (ces), Danish (dan), Dutch (nld), English (eng), Estonian (est), Filipino (fil), Finnish (fin), French (fra), Galician (glg), Georgian (kat), German (deu), Greek (ell), Gujarati (guj), Hausa (hau), Hebrew (heb), Hindi (hin), Hungarian (hun), Icelandic (isl), Indonesian (ind), Irish (gle), Italian (ita), Japanese (jpn), Javanese (jav), Kannada (kan), Kazakh (kaz), Kirghiz (kir), Korean (kor), Latvian (lav), Lingala (lin), Lithuanian (lit), Luxembourgish (ltz), Macedonian (mkd), Malay (msa), Malayalam (mal), Mandarin Chinese (cmn), Marathi (mar), Nepali (nep), Norwegian (nor), Pashto (pus), Persian (fas), Polish (pol), Portuguese (por), Punjabi (pan), Romanian (ron), Russian (rus), Serbian (srp), Sindhi (snd), Slovak (slk), Slovenian (slv), Somali (som), Spanish (spa), Swahili (swa), Swedish (swe), Tamil (tam), Telugu (tel), Thai (tha), Turkish (tur), Ukrainian (ukr), Urdu (urd), Vietnamese (vie), Welsh (cym).
Previously the Multilingual v2 model only supported 29 languages.
Believe it or not, this new v3 model can not only generate voice, but also sound effects ! See here Eleven Labs v3 Prompting guide for full details. Obviously this kind of customization of the audio is not an absolute must for language learning flashcards, but if you spend a lot of time reviewing, it's nice to have a bit of fun with your cards.
[giggles], [laughing], you could add a laugh in the middle of the dialogue.[happy], [excited] will change the emotion in the voice[applause], [clap] also work... for pausesCAPITALS for emphasis.
This is not related to the new Eleven Labs v3 model, and in fact only works with the Eleven Flash 2.5 and Eleven Turbo 2.5 models. It allows you to instruct to the model that your source text is in a particular language, like fr for French. You have to use an ISO 639 language code. This change was contributed by user Wylan on github:
Dutch has a fair number of words that are spelled the same as German or French words (often because they originated in those languages). They are unfortunately pronounced differently though, especially when they contain the letter "g". Using the multilingual models and whatever they do to guess the target language it usually seems to guess German or French. Which to be fair are both much more commonly used language so it does make sense. It does however make it pretty hard to get the correct pronunciation when you're studying Dutch!
There has been a change to the FPT.AI service. Previously, HyperTTS Pro users had to the FPT.AI service. This service unfortunately stopped working for me and I couldn't login, and couldn't add more credits. Their customer support was unresponsive. I realized they also were now offering those same TTS voices on their newer cloud marketplace. So HyperTTS Pro users now have access to this new service, which seems more convenient for international users, and supports international payment options. For those who still want to use the existing FPT.Ai service, you can use the FptAiClassic service in HyperTTS with your own API key. (You will most likely need a Vietnamese payment option such as VNPay) to use it.