You are not logged in. Please login or register.
AudioGames.net Forum → Off-topic room → Is there a multilingual English Spanish speech synthesizer?
I’m looking for a multi lingual English Spanish speech synthesizer that can pronounce words in both languages with no difficulty and a good accent, Especially on the Spanish side. Does such a thing exist? I’m quite surprised that I have never actually come across one.
I know there was a synthesizer that could do this, but I have no idea what it’s called or even if it exists anymore. I am interested to know the answer to this as well.
Are you both talking about the defunct Retorical TTS that was bought by Nuance Communications, INC. And then most of it's administrators moved to CereProc?
I believe that's what it's called, yes. It's too bad it's not being sold anymore, if we do have other multilingual speech engines out there, they're few and far between.
What was very smart of that synth is that each voice speaks they're native language and english as the second. If CereProc want to integrate, or eaven develope, it's technique, that could be awesome.
it's called eloquence, and you're probably using it right now.
No, Eloquence can't transition from one language to another that smoothly. Yes, it's multilingual, but you can't right English and Spanish in the same sentence and expect it to transition from one to the other flawlessly. The voice we're talking about is the Rhetorical speech engine.
Why, in all of that entire time and entire day, decide to revive this topic from almost 5 years ago without prior notice?
What Dan was saying is that Rhetorical Systems, with it's TTS system RVoice, had the capacity of those voices to not only speak their native language, but also speak other languages as well. I believe Nuance Vocalizer has that thing of multilingualism on their voices but still I don't know what it sounds.
According to Rhetorical Systems back in 2004,
Rhetorical wrote:
"Rhetorical's portfolio of text-to-speech (TTS) voices include two Americas Spanish voices and two European Spanish voices.
Technically speaking, the Spanish voices are just like any of our other voices. They can be run on the rVoice 4 engine, on their own or in combination with any of our other TTS voices. They have the same performance as any of our other languages and voices, use the same APIs, and run on the same platforms and operating systems.
Bilingual voices
The male Americas Spanish voice is a bilingual voice, based on a bilingual Spanish speaker living in New York. The European Spanish voices are bilingual voices based on bilingual Spanish speakers living in Spain.
The bilingual voices do automatic language switching: an English sentence will come out in English, a Spanish sentence will come out in Spanish. It also automatically disambiguates words. For example, if you click on Central Bank es un banco central you will hear that the two occurrences of the word "central" are pronounced differently - the first occurrence is pronounced as an English word, the second as Spanish. In addition, XML markup can be used to force the system to switch between languages.
Addresses from North America, Puerto Rico, Mexico, Venezuela and Spain are disambiguated automatically by the Americas Spanish voice:
if you click on 7th Ave. you will hear that it comes out as "Seventh Avenue"; however, C/ Juan Veinte is read out as "Calle Juan Veinte".
The Americas Spanish voice will pronounce North American place names in the style most acceptable there. Placenames of Spanish origin (like Los Angeles are spoken in Spanish; placenames like New York are spoken in English."
One thing that I'm trying to do on Piper tts is like of a bilingual Spanish and English voice model of my self. However I don't know if it treets the way on how the voice can handle both languages and even disambiguate them as RVoice did.
I mean I've discovered that synth back in 2014 I believe, approximately 10 years after Nuance baught the company on november 2004.
That's an interesting topic to discuss, as I don't know if Piper in a future version could handle it.
Now that I am more bilingual myself, I also have a renewed interest in this idea. I am a native English speaker, but my second language is Japanese, which Piper doesn’t support yet. What facinates me is that multi-lingual voice tech has existed in a single form for a long time, and nobody has tried to advance that field until nearly 20 years later.
Me being bilingual myself (l1=Spanish l2=English), I can tell from my view that this technology of a single voice speaking multiple languages instead of switching to another voice for that other language is something that needs re-discovering. Rhetorical Systems's technology of multilingualism was, and still is, unique on that perspective.
Sadly nothing stands or lives forever, as companies and technologies go, we somehow forget it but sometimes we re-discover it.
Someone proposed that idea of adding support for Japanese on Piper.
Indeed, it's the path that new technologies are taking. OpenAI's synthesis engines can switch between languages seamlessly using the same voice. It appears that this will gradually become the standard everywhere, we just have to wait for the entire matter to become democratized.
I am mostly an English monoglot, but if I have one complaint about the default British English voice for espeak-ng, it's how poorly it handles some non-Latin character sets... in particular, it's incapable of stringing greek letters into words, and when reading Japanese, it reads kana as "Japanese Letter" and Kanji as "Chinese Letter", and I'm a big enough geek and a big enough fan of Japanese pop culture that I run into enough Greek and Japanese on otherwise English pages for that to be annoying(Think of all the wiki pages that introduce something by its English name, but have the greek roots of the term or the Japanese name in parenthesis)... I believe there are similar failure modes with Hebrew, Arabic, Cyrillic, and even some of the fancier variants of latin characters, but I run into those less often.
AudioGames.net Forum → Off-topic room → Is there a multilingual English Spanish speech synthesizer?
Generated in 0.023 seconds (54% PHP - 46% DB) with 11 queries
