Digital Speech Fragments

AviaVox passenger announcements sound human, and they’re always 100% grammatically correct. Anyone would think they’d just been read aloud by an articulate human, yet they were generated by very sophisticated software using digital speech fragments.

Did you know that we can generate around 1.8 trillion announcements in every supported language? And since we were founded in 1995, our software is robust and well-maintained. It’s why we’re in more airports than any other provider.

How We Generate Our Announcements

We use voice professionals to record a huge vocabulary of words in each of our supported languages. Our software then splices these words into tiny digital speech fragments, each a very specific sound, and stores them in our database.

When a passenger announcement is generated, AviaVox software re-assembles the appropriate fragments to create the correct words with the right syntax. Thanks to our AI and grammar driver, this is done perfectly – taking into account all the complexities of each language or dialect.

The resulting passenger announcement can’t be distinguished from a native human speaker. It is also clear and well-paced, and doesn’t contain any hesitations or conversation fillers (like ‘um’), or any mistakes (“…Gate 9 – no, sorry, I mean Gate 19”).

Perfect Language Adaptations

Our passenger announcement software can easily handle linguistic sensitivities such as:

• very complex grammar, for example Arabic, Mandarin, Hungarian, Korean and Japanese
• syntax (word order)
• grammatical cases, eg nominative, accusative, genitive
• grammatical genders, eg masculine, feminine, neuter
• social/cultural linguistic norms specific to particular languages or dialects
• linguistic politeness, ie hierarchies of courtesy
• speech dynamics, eg bridging between words, how sounds melt together, how words affect others in proximity
• the way intonation (pitch), emphasis (stress), volume and pauses can affect meaning.