Technology journalist Anicet Mbida unveiled a magnetic smartphone device on French television channel TF1 on Tuesday that translates speech by cloning the user's voice.
The compact accessory snaps to the rear of a mobile handset using magnets and connects directly to a translation smartphone application. Once attached, the hardware delivers spoken translations in another language within seconds, removing foreign accents while preserving the user's natural vocal timbre.
Mbida showcased the translation hardware during his regular segment, C'est déjà demain, on the morning programme Bonjour ! La Matinale TF1 on Tuesday, November 3, 2026. He explained that the unit could significantly ease daily communication with foreign speakers.
During the television broadcast, Mbida said the process was not magic and instead operated through a mobile application running on the smartphone. Users select their preferred language pairing on the phone screen before speaking into the microphone.
Live studio demonstration
To demonstrate the system live in the studio, Mbida began with a test between French and English. After pressing a button on the unit, he spoke a greeting in French, welcoming everyone to the programme and asking those present how they were doing.
Following a very short pause, the device delivered the English translation aloud. Mbida listened to the playback and confirmed to the audience that the synthesized audio matched the natural timbre of his voice.
Mbida then reversed the process, speaking English to receive an immediate French translation. The demonstration continued with a test in Italian, prompting praise from fellow presenters in the studio who remarked that the accent was flawless and that the system worked very well.
The system aims to allow speakers to converse in languages such as English, Italian, or Spanish without worrying about their pronunciation or accent.
Voice cloning process
For the device to reproduce someone's speech in another language, it must first clone their unique vocal characteristics. Mbida explained that a user must read a sample text aloud for roughly 40 seconds so the software can record their vocal profile.
Once that sample is recorded, the system requires about ten seconds to process the acoustic data and generate the cloned voice profile. After that brief calibration step, the hardware is ready to translate spoken phrases into foreign languages.
Voice cloning relies on neural audio synthesis and machine learning models that analyze pitch, resonance, and cadence from audio recordings. These deep learning systems map those personal vocal traits onto the speech sounds of foreign languages, enabling synthetic speech to retain the identity of the original speaker.
TF1 is France's oldest national commercial network and the country's most watched television broadcaster. Mbida is an established science and technology journalist in French media, presenting regular segments on consumer electronics and digital innovation for national television audiences.
Limitations and retail pricing
Although the live broadcast demonstration impressed the studio panel, Mbida cautioned viewers that the technology is not completely infallible. He warned that because the device relies on automated computer translation, small translation glitches can occasionally occur during conversations.
Despite those occasional software errors, Mbida said users should still be able to make themselves understood during routine daily conversations abroad.
The gadget is arriving on the consumer market in two versions. Handheld translation hardware has increasingly shifted from simple digital phrasebooks to artificial intelligence accessories designed to work alongside smartphones.
Depending on the model chosen, the magnetic translation unit costs 90 euros or 150 euros.
