Voice Cloning: How to Speak a Foreign Language in Your Own Voice
The most annoying part of ordinary machine translation isn't the mistakes — it's the voice. Synthesized dubbing sounds faceless, and the other person ends up talking to a budget version of Siri instead of to you. Voice cloning solves exactly that problem: the translation is still a translation, but it sounds like you.
How voice cloning works
The user records a short sample of their own speech once — about 60 seconds. An AI model analyzes the timbre, pace, intonation, and characteristic features of the pronunciation and builds a voice profile from it. From then on, when the system translates your speech into another language, the speech synthesis uses that profile instead of a standard voice — so the translated phrase sounds as if you said it yourself, just in a different language.
How this differs from ordinary text-to-speech
Regular TTS (text-to-speech) draws from a fixed set of standard voices — a male one, a female one, a few variants — and uses them to voice any text, regardless of who actually said it. Voice cloning personalizes that step: the voice is generated for a specific person from their own sample, not picked from a catalog. The difference is especially noticeable on calls: within a few seconds, the other person gets used to recognizing you by voice, even while hearing a translation.
Is it safe
A fair question is whether this turns into a deepfake tool. By design, voice cloning in Loquora only works inside the user's own conversation: the model only voices what you say in real time during a call — it doesn't generate speech on demand for arbitrary typed text. Your voice profile is tied to your account and used to translate your own speech, not to "voice" random text in someone else's voice.
Where it's used
Most often in the same places as voice translation in general: work calls where it matters that a client or partner hears you and not a faceless voiceover; personal conversations with family and friends abroad; and presentations or talks to an international audience, where you want to sound natural rather than like an answering machine.
How to try it
The feature is available on paid plans (Lite+ and up); on the free plan you can use translation as subtitles without dubbing. To turn on cloning: record a voice sample in the app (about a minute of reading any text), pick your working languages — and your voice will be used automatically on every call from then on.