Google announced its brand new feature for Google Meet at Google I/O 2025. Google Meet now features real-time speech translation. Google DeepMind developed a large language model for audio to power the new Gemini-based capability. This allows fluid multilingual conversations by translating the real words into the preferred language of the recipients. However, this also preserves the speaker’s tone of voice and emotional implication.
Live Translation for Google Meet
Google Meet is being used by approximately 300 million people every month as of 2025. The tool has been used for video conferencing, considering both personal and professional contexts. Multiple multilingual Google Meet sessions feature two major types of AI-powered translation options:
AI Speech Translation
The Google I/O 2025 witnessed the launch of newest update on Google Meet- AI speech Translation. This new update on premium Google Meet utilizes AI to provide real-time voice dubbing in another language. Unlike reading subtitles, people can hear a synthetic voice translating the meeting discussions. Currently, this is available in English and Spanish, but it will soon be available in other languages.
AI-powered Captions and Subtitles
Google Meet can automatically produce real-time subtitles of conversations in a conference. These can be seen in the speaker’s language or based on the premium and settings. This can automatically translate the conversation into a different language. For instance, if a speaker speaks in English, people can select to read captions in Japanese. This helps in bridging the linguistic gaps and ensures accessibility of the meetings for everyone.
AI-powered Speech Translation
The new feature is beyond the previous real-time caption translations by enabling audio-to-audio transformations. This means users can hear their counterpart’s voice in the conversation. However, do not confuse this with a robotics substitute; it is a dubbed version in another language with natural-sounding vocal cues. An English and Spanish speaker were able to smoothly communicate in a live demo, with every person hearing a translated version of the other’s speech in their native language. There is a background playback of a faint version of the original voice, which allows additional clarity and context.
Google Meet’s newest update is built on DeepMind’s AudioLM, as stated by Google at Google I/O 2025. It is trained particularly for high-fidelity speech translation. The outcome is low-latency dubbing, which seems to have a live translator in the call. However, Google states that there could be delays due to processing.
To get started with the feature, users can access the ‘Speech translation with Gemini’ option in Meet’s web interface, where they can clarify both the language they are using and the one they want to hear. After enabling this, the interface will show a shining Gemini overlay and a translation metric in the top corner of the screen.
The feature is now available in beta for users on Google’s AI Pro and newly introduced AI Ultra plans. The AI Ultra Plans are available at a cost of $249.99 per month. Only one person in a call needs a subscription to access the feature of speech translation feature. This makes it more feasible for group settings.
At first, the feature is available in English and Spanish. However, it is expected to expand into other languages in the upcoming days. Google at Google I/O 2025 also announced that initial testing for professional users will start lately this year for workspace consumers.
As Microsoft Teams is also experimenting with AI-powered translations, Google’s initiative shows an emerging trend of real-time, AI-powered conversation overcoming language barriers.
Kwon Kim showed how the emerging technology is equipped to capture the tone of the speaker, emotion, and intonation in the translated language. This facilitates free-flowing communication with someone speaking a different language.
If you have ever used Google Translate or another translating tool to communicate with a cab driver in an international location or to order food in a different language, you may have the idea of live translation. Google used the instance of speaking to an Airbnb host regarding the upcoming trip. However, one may imagine communicating with relatives or carrying out research with individuals who speak in diverse languages.
Undoubtedly, there are some drawbacks related to the tool, as we may never need to learn new languages with such updates and cutting-edge technologies. However, it has sufficient potential for communicating key information in a while.
Translated Captions Feature
At Google I/O 2025, the captions feature is also announced by Google. The caption feature of Google Meet automatically transcribes the spoken language during a conference and shows it on the screen in real-time. Based on the plan, captions can be found in the original spoken language or in a translated version.
Live Captions can be enabled by any participant in a meeting or conference. Once activated, Google Meet automatically translates spoken language and displays subtitles on the screen. This can be particularly beneficial for individuals in noisy environments or those who require additional support.
People can enable this new feature by tapping ‘More’, ‘Settings’, ‘Captions’, and then selecting their preferred language. Captions remain at the bottom of the screen and scroll automatically in real-time.
As stated at Google I/O 2025, Google Meet enables Business Plus, Education Plus, Gemini-enabled editions, and Enterprise Standard users to translate the captions into other languages. The spoken language is tracked with a simple activation, and the subtitles are automatically displayed in the preferred language.
For instance, a speaker using the English language can be understood by an individual reading subtitles in Japanese. This is because of the real-time use of AI-powered translation.
As of 2025, the captions feature more than 80 languages. However, some are still in beta and may be unavailable to users. Subtitles support more than 4600 languages for the people who have access to Gemini for Google Workspace or other eligible plans.
To enhance readability, Google Meet also enables people to customize the appearance of the captions. Users can now change the font size, color of text, and background to address their accessibility preferences. Such styling updates became available across the world.
Limitations of Google Meet’s Real-time Translation
Although Google Meet has made significant strides with AI-powered live translation as discussed at Google I/O 2025, there are some potential limitations associated with it.
Google Meet does not feature Remote Simultaneous Interpretation (RSI) channels like specialised meeting platforms. This means it does not offer individual audio channels for interpreters and listeners. There is no default way to connect human interpreters ot a meeting or offer several audio channels for people to choose from. This can be challenging for hosts who require live interpretation from professional experts in multiple languages.
For professional interpretation, where human beings listen and speak in real-time. Several firms utilize third-party channels, such as Interprefy, in conjunction with Google Meet. Such platforms connect with Google Meet to offer a separate audio feed for interpreters as well as attendees. However, this is not a default feature of Google Meet.
Google Meet’s latest AI speech translation, introduced at Google I/O 2025, rolled out noticeable improvements in live language accessibility. However, it also has some limitations.
Firstly, language support is limited since real-time speech translation is presently available in English and Spanish. Google has launched plans to expand the support to German, Portuguese, and Italian speakers in the upcoming days. Even though the features like translated captions support more than 69 languages, the spoken AI translation feature stays limited when it comes to language pairs.
Again, access to Google Meet is limited to users who have bought plans for Google Workspace AI Pro and AI Ultra Plans. This is not a free plan.
Summary
At Google I/O 2025, Google announced the newest update on Google Meet, which has become a platform for global communication. The latest updates, such as AI speech translation, help in surpassing what is possible in different languages. However, there are certain limitations that you must consider.
Also Read:
