Guide
AI voice synthesis (text-to-speech)
Voice synthesis, or text-to-speech, is the technology that turns written text into audible speech. It has existed for decades in robotic, choppy forms; recent models produce a voice with a pace and intonation close to human speech, which fundamentally changes what you can do with it on the phone.
What changed with recent models
A classic voice generator reads text in a linear way, at a constant pace with little nuance. A recent voice synthesis engine modulates intonation based on the meaning of the sentence, places natural pauses, and can even reproduce slight hesitations. Callifly uses ElevenLabs as its voice synthesis engine, which covers over 70 languages natively, including French, English, Spanish, Arabic, German, Japanese and Portuguese.
Free voice generator vs. phone-call voice synthesis: the difference
A free online voice generator converts text into an audio file to listen to or download: useful for a voiceover or a recorded message, with no real-time constraint. The voice synthesis used in a phone call has to produce speech instantly, in response to what the caller says, with low enough latency to not break the conversation's rhythm. It isn't prepared text: the sentence to speak is generated by the understanding engine during the call, then converted to voice on the fly.
What voice synthesis doesn't do on its own
Voice synthesis doesn't understand what the caller says: that's the job of speech recognition and the understanding engine, separate building blocks. It also doesn't decide what to answer: it voices the response already determined. For how a full call works, see AI voice agent, which details how these pieces work together.
Voice customization
The vocal style (tone, pace, voice gender) is chosen to fit your brand and industry. A real-estate agency and a technical support line don't necessarily need the same register. This setting is configured once, when the agent is set up.
Frequently asked questions
Can you use Callifly just for voice synthesis, without a full agent? No, Callifly offers a complete voice agent (understanding, business data, voice synthesis) rather than a standalone text-to-speech service.
Can the voice switch languages mid-call? Language detection is automatic at the start of the call; switching languages mid-conversation depends on the scenario defined with you.
How long does it take to set up a custom voice? Choosing the vocal style is part of the agent's initial setup, taking a few minutes.