If Roblox were to implement this, the use would be having text-to-speech without every developer having to upload 44 different sounds (per voice) and script a system for it. I can’t imagine splicing sounds like that would sound nearly as good either, since text-to-speech algorithms are much more complicated than that.
PLEASE ADD THIS OMG because of the audio limit it is so hard to make characters talk.
Upload 7-minute-long segments of dialog into one audio. Roblox will release a feature soon allowing you to play certain segments of audio at a time.
Trust me, I read the documentation.
This would be really useful for games with a lot of dialog. You could use that for that
It would be easier still to just have a command like
tts(script.parent.text)
If they actually made that command I am pretty sure everyone will be happy.
One simple way for roblox to implement this both client side and server side is using Mimic 3 by Mycron AI. Its a open source TTS implementation, with hundreds of voices in a lot of languages, most of them sounding very realistic, with a simple web API. Roblox just has to integrate it into the player client, and create a simple wrapper for the http api. Something like this would be nice:
local audioResult = tts("Text to be spoken", {
"voice": "en_US/cmu-arctic_low", // voice key
"speaker": "fem", // default speaker
"length_scale": 1.0, // speaking rate
"noise_scale": 0.667, // speaking variablility
"noise_w": 1.0 // phoneme duration variablility
})
It should return some id (For example rbxttsaudio://12345678) That will be disposed in a few minutes, so we can use it in actual audio objects.
Pronunciation goes brrrrrrrrrrr
Can you give a script to Implement this using http api?
There’s no way to implement this in roblox, since there will be no way to actually play the sound (you cannot upload audios in real time, and there is a big delay for uploading audios, which is why I want this to be a roblox feature)
Bump, I would love this being added, like there are so many things developers could do with this, for instance, when a player wins the game say it in text to speech.
Player.Name.." won the game with"..points.." amount of points"
Someone at ROBLOX made a prototype during hack week of 2018, this needs to be added please.
Also bumping this, if roblox is looking towards accessibility then they have to finally consider t2s, it’s way too complicated and low-quality when it’s a custom system due to the limit of audio uploading and the time it takes for it to be moderated + lua is slow as hell. We have voice chat so why not this? Especially since there’s “AI” generated voices nowadays and Roblox is investing in this subject I can imagine that practically all games would benefit from this or use it lol
Support. Would be useful for accessibility as well as games with captioning or cutscenes.
I see this probably better than uploading audio assets, since it would save my quota from being used.
One drawback, moderation, some developers could take advantage of this and use it to break the roblox ToS (profanity, and saying to visit certain inappropriate sites), so there would have to be some sort of restriction on what can be said, so for example, to replace or refuse to play the audio.
Just make the place 18+ only until the speech is verified by a moderator or something. I would say filter but that’s even worse.
I’d love to see this feature implemented. I’m thinking of adding an accessibility feature for my game, similar to Mac OS X and iOS’s VoiceOver, but I can’t get it to work easily thanks to audio upload limits and well, Roblox engine limitations.
Having the native ability to use text-to-speech voices through a service (for example, “SpeechService”), with the ability for using custom pronunciations, etc, would be a major stride for accessibility and fun gimmicks in experiences.
It would also be awesome for the ability to add custom voices for text to speech as well, but I don’t think that would be easy to implement at all.
For the properties, it could be Enum and stuff, like tts("text", Enum.VoiceLanguage.English.US, Enum.VoiceSpeaker.ICarly, 1.0, 0.667, 1.0)
Bumping time: I’m pretty sure this is going to happen soon
Considering these changes from yesterday and today:
AudioTextToSpeech | Roblox API Reference (0.664.0.6640713)
+ [DFLog::TextToSpeech] SoundService::loadText won't work if the cookie is empty or invalid
+ [DFLog::TextToSpeech] SoundService::loadText request: {}
+ [DFLog::TextToSpeech] url={}
+ [DFLog::TextToSpeech] SoundService::loadText callback: length is zero, {}
+ [DFLog::TextToSpeech] SoundService::loadText : request failed {}
+ [DFLog::TextToSpeech] baseUrl = {}
its not implemented YET, but maybe it’ll be implemented just around patch 665 or 666
End of June