
DeepL Voice
: How to use it, features, and the business problems it solves
What is DeepL Voice?
DeepL Voice is the real-time voice translation solution from DeepL SE, the company behind the DeepL translator. It works with Microsoft Teams, Zoom Meetings, and Google Meet, showing captions in over 40 languages — a different language per participant within the same meeting. Speech-to-speech translation that keeps the speaker's voice has been generally available since September 2026, and in November 2026 DeepL added Aure, a voice model that carries changes in tone and emotion into the translated voice (eight languages including Japanese). Transcripts and translations are never stored permanently and are deleted once the call ends. Pricing starts at ¥2,200 per user per month for Voice Starter (tax included; ¥1,900/month billed annually), with a 14-day free trial.
Business problems it solves
About "DeepL Voice"
What is DeepL Voice
DeepL Voice is the real-time voice translation solution from DeepL SE, the company behind the DeepL translator. Separate from the classic DeepL that translates documents and text, its purpose is to turn spoken words into captions as they are said.
It launched on 13 November 2024 in 13 languages including Japanese, expanded in July 2025 to 16 speech input languages and 35 caption languages, and reached over 40 languages in April 2026. On 15 September 2026 speech-to-speech translation that keeps the speaker's own voice characteristics became generally available, and on 11 November DeepL added Aure, a new voice AI model that carries shifts in tone and emotion into the translated voice, in eight languages including Japanese. Its distinct position among meeting translation tools is that an organization already standardized on DeepL can extend the same translation quality and the same security baseline to its meetings.
DeepL Voice is not one product
Three solutions ship under the DeepL Voice name. The first is the one used in meetings.
01DeepL Voice for Online Meetings
Real-time voice translation for web conferences — the focus of this page.
- Platforms — Microsoft Teams, Zoom Meetings, Google Meet
- Available now — real-time captions in over 40 languages, speech-to-speech translation (generally available since 15 September 2026)
- Desktop app — for Windows and Mac (Intel Macs not supported). Detects meetings automatically and overlays captions on top of whatever you are working on
02DeepL Voice for In-Person Conversations
For conversations with someone in the same room — a meeting room, a shop floor.
- Environments — iOS, Android, web
- Scope — one-to-one and group conversations
- Processing — supports on-device voice translation
- Speech-to-speech — supported (Aure applies automatically)
03DeepL Voice API
For embedding voice translation into your own product or contact centre.
- Typical uses — multilingual customer service and sales handling
- Capability — voice translation and real-time interpretation delivered as an API (Aure applies automatically)
Speech-to-speech translation and the new Aure model
DeepL Voice is not limited to captions — you can also listen to the translation as speech. This feature used to be listed as "coming soon"; on 15 September 2026 it became generally available for online meetings, in-person conversations, and the API. On the pricing page, "Voice-to-Voice translation" is included from the entry-level Voice Starter plan.
| Date | What changed (DeepL announcements) |
|---|---|
| 15 September 2026 | Speech-to-speech translation became generally available. In 14 languages including Japanese, the translated speech keeps the speaker's own voice characteristics |
| 11 November 2026 | New voice AI model Aure announced. It carries changes in tone, emotion, emphasis, and pace during the conversation into the translated voice in real time |
Aure builds on the September model. The earlier model generated the translated voice from the voice it captured at the start of the conversation; Aure keeps analysing speech throughout, so the translated voice reflects not only "who is speaking" but also "how they are speaking" as it changes. DeepL's examples include adding urgency by stressing words in a negotiation, or shifting into an enthusiastic tone when sharing good news.
Sources: DeepL press release on Aure (11 November 2026), Aure product page, announcement of speech-to-speech availability (15 September 2026)
One meeting, a different caption language for each participant
DeepL Voice for Online Meetings translates multiple languages of audio within a single meeting. An utterance spoken in English appears simultaneously in German on the screen of a participant who chose German, and in Japanese for one who chose Japanese.
Captions carry the speaker's name, so in a meeting with several speakers you can still follow who said what. Each participant speaks and reads in their own language; nobody has to converge on English as a shared tongue.

Source: DeepL blog, "Introducing DeepL Voice: real-time voice translation" (subsequent images are from the same source)
The caption target language is also not limited to the languages supported for speech input. The official FAQ explains that while only certain languages are supported for voice input, captions can be translated into any language available in DeepL translation. The listener's side of the matrix is wider than the speaker's.
A DeepL bot joins the meeting
The thing to check before you commit is how audio is captured. DeepL Voice for Online Meetings uses a DeepL bot that joins the meeting and receives the audio there.
The official help centre states that for the bot to join, the organization must permit external participants in meetings; on Microsoft Teams, chat with external and anonymous users also has to be enabled. For organizations that block external guests from joining, changing that setting is a precondition for adoption.

If you cannot put a bot in the meeting, look at tools that capture browser tab audio or microphone input directly — [Sentio](https://ai-best-search.com/products/sentio), or the PC audio translation mode in [VoicePing](https://ai-best-search.com/products/voiceping). When external sales calls are the main use case, this difference matters before either price or accuracy does.
How to use it
-
Choose a plan
Pick Voice Starter or Voice Business on the Voice tab of the pricing page. Voice Starter can be tried free for 14 days. Teams of 50 or more contact sales for custom pricing.
-
An administrator installs the app
An administrator adds the DeepL Voice app to Microsoft Teams, Zoom Meetings, or Google Meet, and confirms that external participants are permitted to join meetings.
-
Assign licences
Assign DeepL Voice for Meetings licences to the people who will use it.
-
Turn on captions in the meeting
Launching DeepL Voice during a meeting brings the DeepL bot in and starts real-time captions. Each participant picks their own display language. To hear the translation as speech, turn on voice translation.
-
Keep the transcript if you need it
Since the July 2025 update, the full meeting transcript and its translation can be downloaded.
Pricing
Pricing is published on the official pricing page (the "Voice" tab of the DeepL Pro pricing page). All prices are per user per month and include tax (as of 20 November 2026).
| Item | Detail (official wording as of 20 November 2026) |
|---|---|
| Voice Starter | ¥2,200 billed monthly / ¥1,900 billed annually (monthly equivalent). 300 minutes per user per month, real-time translated captions, speech-to-speech translation, transcript download |
| Voice Business | ¥6,200 billed monthly / ¥5,280 billed annually (monthly equivalent). 600 minutes per user per month (shared across the team), SSO and SCIM |
| Large teams (50+ users) | Custom pricing (purchase order and invoice billing, dedicated contact) |
| Setup fee | Not listed on the pricing page |
| Free trial | Voice Starter is free for 14 days (Voice Business has no free trial and starts on purchase) |
As of 20 November 2026 the pricing page showed a discount: 25% off the first year on annual billing (Voice Starter ¥1,425 and Voice Business ¥3,960 as a monthly equivalent) and 25% off the first month on monthly billing (¥1,650 and ¥4,650). No end date is given, so check the pricing page when you sign up. The table above shows the regular, undiscounted prices.
Annual prices are shown as a monthly equivalent. The annual discount is about 14%, unlike the 16% annual discount on the DeepL translator plans. Voice pricing sits behind a tab on the same pricing page as the translator, so it is easy to miss, and the display currency changes with the region you access it from.
It used to be quote-only, but like Sentio and VoicePing it now publishes a price list, so you can get a sense of cost early in evaluation. The DeepL translator's Individual, Team, and Business plans also include 30 minutes of Voice per month.
Language coverage
Real-time captions support over 40 languages, including Japanese, all 24 official EU languages, Vietnamese, Thai, Arabic, Norwegian, Hebrew, Bengali, and Tagalog. As noted above, the set of languages supported for voice input differs from the set available as caption targets.
Speech-to-speech translation works in over 30 languages. Of these, 14 — including Japanese, English, Chinese, Korean, German, French, and Spanish — keep the speaker's voice characteristics (called "Speaker match" in the official help); the others are read out by a third-party voice that does not match the speaker (official help: DeepL Voice languages). Aure, which also carries changes in tone and emotion, covers eight of them.
Security and data handling
The official FAQ answers the question that matters most for enterprise use directly.
If you want to accumulate meeting logs and search them later, this "keep nothing" design becomes a constraint instead. When building an archive of minutes is the goal, consider a tool that retains meeting logs such as VoicePing, or pair DeepL Voice with an AI meeting-minutes tool.
Deployments and reputation
More than 200,000 companies use DeepL overall. In Japan, total DeepL Voice meeting time in July–September 2026 was more than seven times that of the same period a year earlier, and the number of adopting companies in Japan and Korea grew about 3.5 times, DeepL announced on 16 November 2026 (press release). Japanese users it highlights include NEC (since December 2024, with Microsoft Teams), Asahi Group Holdings, and Miyazaki Prefectural Government (for Japanese transcription). Two overseas customers are published as case studies.
01Inetum
A leading European IT consultancy with 28,000 employees.
- How they use it — DeepL Voice combined with the API
- Outcome — faster knowledge sharing, and the ability to staff people by skill rather than by language ability, as published
02Aramark / Avendra International
A major food service and facilities management group and its procurement arm.
- Outcome — adopting DeepL Voice cut global meeting time by 50%, as published
For third-party assessment, DeepL publishes a comparison by the translation industry publication Slator, pitting DeepL Voice against Microsoft Teams, Google Meet, and Zoom: a translation quality score of 96.4 out of 100 (competing platforms scored 87–89), an error rate of 4% against a 17% average, and 96% of language experts placing DeepL Voice first in a blind test.
These figures come from a report DeepL publishes on its own site. The assessment was carried out by Slator, a third party, but DeepL is the one publishing it — read it with that in mind. Real-world accuracy will vary with the jargon used in your meetings and the accents of your speakers. Test it on your own material during the free trial.
How it differs from other AI interpretation tools
| Product | Audio capture | Spoken output | Pricing |
|---|---|---|---|
| DeepL Voice | A DeepL bot joins the meeting | Available (keeps the speaker's voice in Japanese; supports Aure) | Flat rate (from ¥2,200/month, tax included) |
| Sentio | Browser tab audio / microphone | Available | Flat rate (from ¥3,300/month) |
| VoicePing | Both PC audio and a meeting bot | Available | Metered plus flat rate |
| kotoba | Smartphone app | Available | Flat rate (5 free hours/month) |
| Wordly | Meeting and event platform integration | Available | Metered by time |
DeepL Voice suits organizations already standardized on DeepL that want one consistent translation quality and security baseline extended into meetings. It does not suit environments where a bot cannot join. Spoken output has been available since September 2026, and since November Aure also conveys changes in how you speak. If document and web page translation is the real goal, see the page for DeepL itself.
Information on this page is based on what the official site published as of 20 November 2026 (case studies and third-party evaluations as of August 2026). Please check the official site for current specifications and pricing.


