Editing the order of voice script

Hi, new user, installing a new server.
After using some TTS voice service to translate some of the voice files from English to Romanian, I need to rearrange the order that the voice files are played (in voice mail) because of grammar. Is there a config file with the scripts?
Just translating the voice does not really work, is sounds weird not using proper syntax or grammar.

Thank you!

I don’t know how FreePBX handles say.conf.

In the soiurce of the english sounds


will be a files named core-sounds-en.txt which contains the english text of the words spoken in the named files, possibly a good reference to start the process.

Already done those (about 200 for now), I need to reorder the voice prompts.

In English is - You have no new messages.
In Romanian is - You no have messages new. (spoken in Romanian of course)

Just an example, not the exact message. And I have to add extra sound to have proper grammar. Modifying a sound to fit a certain message could resoult making no sense in other messages, so that’s not an option.
These are just tweaks for the internal extensions, outside people will only hear prerecorded IVR menu.

So are you sending “You no have messages new " in English or a Romanian translation of “You have no new messages” to your TTS engine”

no, I translated using TTS the .wav files

I’m not at work but, I remember a script

vm-youhave.wav + vm-no.wav + vm-INBOX.wav + vm-messages.wav
You have + no + new + messages.
And in Romanian is
“Aveți + nu + nou + mesaje.” is completely wrong and should be “Nu aveți mesaje noi” 3 of the 4 sound I have but are not in the right order and I have to add a variation to vm-INBOX.wav to decline the word NEW, I cannot record the “noi” for vm-INBOX.wav, because it may break sentences were the word “nou” is correct, so I need 2 versions of vm-INBOX.wav (vm-INBOX1.wav vm-INBOX2.wav) depending on what script is read. That is what I’m looking for, a config file that I can edit the script, maybe a list of all hard-coded voice messages may be helpful.

you can’t TTS a wav file, you can only STT one but syntactically you will need to feed text in the language you want to transpose, maybe soon AI will fill that gap for you but right now separate the two tasks, so what does your TTS engine make of nu ai mesaje noi which is a google translation of “You have no new messages"

TTS doesn’t make sense. You would have to do TTS back to back with STT, but that would be equivalent to doing spoken language translation which would have no explicit text intermediate.

I don’t think there is a solution to the word order issue, but it seems to me that “new” is really a shorthand for the noun phrase “new message mailbox”, which you can get away with in English, because adjectives don’t decline. As a lot of languages decline adjectives, I would have thought there were tactics for this in more commonly used Asterisk languages. I would assume that Spanish has the declination issue, and that is heavily used. Italian also seems to be fairly commonly used. I don’t know if people just tolerate it, or if they use noun phrases that are safe.

Only knowing that Romanian is in the same family as French, I wonder if your attempted Romanian translation of “no” is negating the sentence. The English meaning of “no” here is “zero”, although using “zero” in English would be pretentious.

This comes back to an extension of the first point. Any translation needs to start by translating the whole sentence, not the component phrases. If you translate “no”, on its own, the translation is likely to translate it as the opposite of “yes”.


Likely burgeoning AI techniques will help twice here. Current engines do better with English and German than Romanian or Lithuanian for obvious reasons.

How does sending to the TTS

demo-instruct: Dacă doriți să aflați mai multe informații tehnice despre Asterisk dial 2 acum. Dacă doriți să testați capabilitățile Voice over IP ale Asterisk, puteți forma 500 pentru a încerca o conexiune Inter-Asterisk Exchange sau IAX la un server demonstrativ la Digium. Pentru ca acest test să funcționeze, va trebui să fiți conectat la Internet și să aveți cel puțin un modem de 28,8 kilobiți. Pentru a executa un test de eco, formați 600. Acest test este cel mai util atunci când v-ați conectat la acest server Asterisk dintr-o locație de la distanță. Configurația eșantion are, de asemenea, un singur utilizator cu extensia 1234 și parola 4242. Acest utilizator este configurat să sune pe consola atunci când este apelat extensia sa. Dacă apelați 1234, puteți încerca să suni pe consolă. Dacă consola este ocupată sau indisponibilă, vi se va oferi opțiunea de a părăsi mesageria vocală. Pentru a verifica mesageria vocală pentru utilizator, apelați extensia 8500 pentru a intra în sistemul de mesagerie vocală. În cele din urmă, puteți apăsa tasta pentru a vă deconecta de la PBX.

workout ?

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.