With AI-supported translators, the manufacturer Timekettle repeatedly caused surprised faces at its own stand at the IFA. The new “W4 Pro AI Translator Earbuds” significantly reduce the latency when translating and come in a rather unusual charging case. In an interview with nextpit, Timekettle CEO Li Tian reveals how the new earbuds are supposed to ensure natural conversations.
In the last few months I have tested two devices from Timekettle for nextpit: the Timekettle X1 (for testing) and the Fluentalk T1 Mini (for testing). The manufacturer has specialized in translation hardware and is attracting international attention. The speed and quality of the translations is higher than with many competitors. At the same time, conversations with Timekettle devices feel somehow “different” than with a translation app.
Note: This article was created as part of a cooperation with the manufacturer Timekettle. This had no influence on the exact content.
And translation apps are also the first reaction I hear from my colleagues at nextpit and from friends when I tell them about my Timekettle tests. Why not just use an app – after all, they are free and offer really good translations thanks to AI integration. And that was exactly the question I asked Li Tian from Timekettle first.
“Why not just use a translation app?”
In fact, Timekettle also offers an app that can be used to carry out translations. The “Timekettle” application, which was actually intended for use with translation headphones such as the W4 AI Translator Earbuds presented at IFA 2024, also offers a translation function. What is missing there, however, is simultaneous translation, and according to Li Tian, there is a good reason for this.
“We [Menschen] need real translations that don’t disrupt contact and body language. We want to remove the technical device from the conversation as effectively as possible. Translation apps are suitable for translating a word or a sentence – but we are aiming our devices at longer conversations.”
The W4 Pro AI Translator Earbuds, for example, can be used in pairs. One person puts the left earbud in their ear and the other person puts the right earbud in their ear and the conversation is translated mutually and simultaneously. However, CEO Tian emphasizes another aspect to ensure the naturalness of conversations: “Back then, translations were consecutive, but what we humans want is simultaneous translations.”
Timekettle CEO Li Tian invited us to talk at IFA 2024. / © nextpit
Timekettle’s simultaneous translations have reached a new milestone thanks to semantic language processing in the new W4 Pro AI Translator Earbuds. The AI translators divide spoken sentences based on their meaning and context, translating certain segments before the sentence is finished.
“Often you have to [bei Übersetzungs-Geräten] wait for translations. With our new devices, however, you can hear the translation while the other person is talking. Of course, there is always a delay and a latency. But the shorter these are, the more natural the conversations.” Technically, Timekettle has reduced the processing times for translations even further.
“Keyword data protection: How does Timekettle’s triple security layer work?”
Devices like the Timekettle X1 and the W4 Translator Earbuds are designed primarily for companies. They sometimes have sensitive information that naturally needs to be protected during digital translation. In my reviews, I’ve always had to mention that Timekettle offers a triple data protection system, but I’ve never been able to explain exactly how it works. So, Mr. Tian?
“The first rule is: We process the data where it is recorded. This means: If a German Timekettle user translates a conversation, the data is not processed on our Chinese servers. Of course, we follow the requirements of the GDPR and the CCPA certificates in the USA. And to further improve the quality of our data protection, we have commissioned an independent organization to review our data protection guidelines.”
In addition, Timekettle encrypts the data using end-to-end encryption so that it cannot be compromised on the way to the server. The third important aspect, according to Li Tian, is that usage data is not stored when you log in to Timekettle. Users’ own language is so sensitive that it never leaves the Timekettle device when it is in use.
The upgrade to HybridComm 3.0 is expected to improve the speed of translations by 200%. / © Timekettle
If a user wants to use functions such as the new AI-supported voice cloning, which outputs translations in a replica of their own language, they must first explicitly allow this. Only then will Timekettle process the voice recordings.
Exciting: Offline use and the translation of media
Towards the end of our conversation, we discussed the possibilities of the new W4 Pro Translator Earbuds in more detail. And with media translation, they offer an exciting function that I really wanted to try out after the interview. The headphones can translate all audio content that iPhones and Android devices output via the Bluetooth connection.
If I’m watching a video on YouTube and am too lazy to read the automatically translated subtitles, the earbuds can easily translate the information for me. All you need is the Timekettle app and an internet connection – which brings me to the last point of our interview.
Devices like the X1 Translator have an integrated eSIM that works internationally. / © nextpit
Timekettle always offers a limited selection of offline translations on its devices. Between 13 and 16 language pairs can be used even when there is no internet connection. But why can’t you freely decide which languages you want to process directly on the device?
“If you scale down the translation model for offline use, the quality of the translations suffers. The translations therefore become worse when calculated offline. Our next goal, however, is to soon be able to offer more than 20 languages and, in the distant future, all language pairs for use directly on the device.” As an alternative to offline translations, Timekettle integrates an international SIM card into its most expensive X1 device. With this, users can translate anywhere in the world for a year as standard.
Conclusion: Technology should finally no longer stand between us
I think that Timekettle’s attempt to make its own translation devices as invisible as possible in conversations is a very exciting approach. The idea of ”ambient computing”, which the company “Nothing” pursued from the beginning, was still evident at the IFA, even if the initial hype has died down. But ideas like AI-supported video glasses that can display information very discreetly in our field of vision are still interesting.
Technical devices such as smartphones or televisions are particularly disruptive in conversations. We are all familiar with this from situations in which friends or relatives simply cannot put their mobile phones down. And this is where I see the greatest advantage of Timekettle’s devices. At the stand at IFA, it did not take long for Johanna and Eileen to get used to using the W4 Pro and then to have a somewhat natural conversation with a Timekettle employee.
With ever shorter translation times and the semantic division of sentences, Timekettle is improving the “invisibility” of its devices even further. At the same time, AI systems are making significant progress in the replication of voices, and Timekettle is already integrating a “voice cloning” feature into its devices as a test version. So I’m excited to see whether Li Tian and I will be able to communicate in our respective native languages and in our own voices at the next IFA.
And then I will certainly find out whether he has actually tried a kebab, as I recommended after he asked about local dishes – thanks for the conversation!
