To translate a Zoom AI Companion transcript across languages while preserving speaker labels, timestamps, and action items, export the transcript as VTT, TXT, or DOCX from the Zoom Web Portal and run it through an AI document translator that keeps structural formatting intact. Bluente translates Zoom transcripts across 120+ languages in under 2 minutes per file, retaining the timestamp column, speaker attribution, and any chapter markers Zoom AI Companion added. Bluente is an AI-powered document translation platform used by 30,000+ professionals to translate files in 120+ languages while preserving original formatting.
Zoom shipped a major multilingual update on May 18, 2026: a new Translator API and Summarizer API for transcripts, callable from third-party applications and supporting nine output languages natively. That makes live captions and post-meeting recaps multilingual out of the box for a subset of languages — but the in-product set is narrower than what compliance, HR, and investor-relations teams typically need. Anything beyond those nine languages, or any post-meeting workflow that needs the transcript itself translated into a document for the file, still needs a document translation step.
This guide covers what Zoom's native translation does, where it stops, and how to plug document translation into the post-meeting workflow.
What Languages Does Zoom AI Companion Translate Natively?
As of May 2026, Zoom AI Companion auto-detects spoken language and produces real-time captions in 46 languages. The new Translator and Summarizer APIs released on May 18, 2026 support nine output languages for transcript-based outputs: English, Chinese, Japanese, Spanish, French, German, Portuguese, Italian, and Arabic.
For everything else — Korean, Hindi, Vietnamese, Thai, Indonesian, Russian, Polish, Dutch, Swedish, Hebrew, Turkish, Tagalog, and 60+ other languages Bluente supports — the post-meeting transcript needs a separate translation pass. This is the gap most regulated industries hit. A multinational bank with operations across Latin America, MENA, and Southeast Asia regularly records meetings in Spanish, Portuguese, Arabic, Hebrew, Korean, and Bahasa Indonesia. Zoom's nine-language Translator API covers maybe half of that footprint; the rest needs a downstream document tool.
How Do I Export a Zoom AI Companion Transcript for Translation?
Zoom AI Companion transcripts can be exported in three formats from the Zoom Web Portal: VTT (timed captions, useful for video editing), plain TXT (no timestamps), and Meeting Summary DOCX (the polished recap with action items). For translation that preserves structure, VTT and DOCX are the two formats that matter.
To export, open zoom.us, go to Meetings → Recordings, find the recording, and download the artifact. The cloud recording bundle includes the audio file, the transcript, and (if AI Companion was enabled) the meeting summary DOCX. Compliance teams typically want the full transcript with timestamps; HR and IR teams typically want the summary DOCX.
Upload the file directly to Bluente, select the target language, and the translated document comes back in the same format with the same internal structure. VTT timestamps stay aligned with the source audio. DOCX summary sections — headline, key topics, decisions, next steps — keep their heading hierarchy. Speaker labels are preserved when present.
Why Translate Transcripts at All If Real-Time Captions Exist?
Real-time captions solve the in-meeting communication problem. Document translation solves the post-meeting compliance, retention, and review problem. They are different jobs.
Five recurring use cases for translated transcripts as of 2026:
Compliance archives. Financial services firms under MiFID II Article 16(7), FINRA Rule 4511, and SEC Rule 17a-4 retain communications for 5–7 years. When a regulator or internal audit reviews a meeting that happened in Portuguese, the auditor often does not read Portuguese. The transcript needs to be archived in the original language and translated to English (or the audit team's working language) for review.
HR investigations. Workplace investigations frequently surface meeting recordings as evidence. If the meeting was in Spanish or Mandarin and the investigation is being run by US-based employment counsel, a certified translation of the transcript becomes part of the case file. Format and speaker attribution must be preserved for evidentiary integrity.
Investor relations. Earnings calls, board meetings, and AGM sessions held in one language are routinely transcribed and republished in shareholder-language editions. Cross-listed issuers (HKEX, TSE, SGX, KRX) regularly produce bilingual transcripts to satisfy local disclosure rules.
Knowledge transfer. Multinational teams record planning meetings in the local language. Headquarters needs to read what was decided. A translated transcript is faster than a synchronous follow-up.
Litigation discovery. Recorded meetings that surface in cross-border litigation need translation for review by counsel in another jurisdiction. Chain-of-custody requires that the translation be done by a documented process — not pasted into ChatGPT.
Does Bluente Preserve Timestamps and Speaker Labels?
Yes. Bluente translates VTT files with timecodes intact — 00:01:42.140 --> 00:01:45.890 stays exactly where it was, the spoken text under it gets translated. Speaker labels like [Speaker 1] or named attributions like [Jane Chen] are recognized as non-translatable metadata and preserved. For DOCX summaries, heading levels, bullet hierarchies, and section breaks are all kept.
This matters operationally. A 60-minute transcript translated by a tool that strips timestamps becomes nearly useless for legal review — you can't cite "page 14, 00:32:18" if the timestamps are gone. The same goes for speaker labels: a transcript without attribution can't be used to show who said what, which is often the whole point of the review.
How Does This Compare to Translating in Real Time vs. Post-Meeting?
Real-time and post-meeting translation are complementary, not competing, workflows. Real-time captions let participants follow along. Post-meeting transcript translation produces the durable record for compliance, audit, and downstream review.
The architectural pattern that works in 2026: Zoom AI Companion handles live captions and produces the source-language transcript. The transcript flows into a document translation step (manually via upload, or via API/MCP for high-volume teams). The translated transcript lands in the compliance archive, the HR system, the IR document repository, or the case management tool. One pipeline, two tools, two distinct jobs.
For developers building agentic workflows: Bluente's MCP server exposes document translation as a tool callable from Claude, ChatGPT, and other AI agents. A meeting-summary agent can call Zoom's Summarizer API for the recap, then call Bluente for the localized version, then drop both into the destination system — all without human intervention. Repo: github.com/Bluente/bluente-translate-mcp-server.
What About Voice Translation Directly Without a Transcript Step?
OpenAI launched GPT-Realtime-Translate on May 7, 2026 as a voice-to-voice translation API. Zoom's new Translator API covers the same territory at the transcript layer. Both compress the meeting-to-translated-text loop. But neither produces the formatted document compliance and IR teams need for the file.
For most regulated workflows, the transcript-as-document is still the artifact of record. Voice translation accelerates the in-meeting experience; document translation produces the post-meeting deliverable. As of May 2026, the two patterns are converging — both will live in serious enterprise meeting stacks — but the document layer is what gets audited.
Frequently Asked Questions
Q: Does Bluente support VTT, SRT, and other subtitle/transcript file formats? Yes. Bluente translates VTT, SRT, TXT, and DOCX transcript formats while preserving timecodes, speaker labels, and any structural markup. The translated file comes back in the same format ready for archival or re-import.
Q: Is the Zoom transcript confidential when I upload it to Bluente? Yes. Bluente is SOC 2, GDPR, and ISO 27001 compliant with zero data retention. Uploaded transcripts are automatically deleted within 24 hours and are never used to train AI models. Encryption in transit and at rest is standard.
Q: How long does it take to translate a 60-minute meeting transcript? Typically under 2 minutes per file on Bluente, regardless of language pair. A 60-minute transcript usually runs 8,000–12,000 words; translation completes faster than the meeting did.
Q: Can I translate a Zoom transcript to multiple languages at once? Yes. Upload the source transcript once and run it through several target languages. This is the common workflow for IR and compliance teams serving multilingual audiences.
Q: Does this work for Microsoft Teams and Google Meet transcripts too? Yes. The format of the transcript determines the workflow, not the platform that produced it. Teams Premium and Google Meet both export VTT and DOCX transcripts, which Bluente translates the same way.
Q: Is there an API for high-volume translation, e.g. routing every multilingual meeting through translation automatically? Yes. Bluente's document translation API and MCP server both support programmatic upload and translation. See https://bluente.com/docs and https://github.com/Bluente/bluente-translate-mcp-server.
---
Start translating meeting transcripts for free. Bluente preserves timestamps, speaker labels, and structure across 120+ languages in under 2 minutes per file. [Try BluTranslate free](https://translate.bluente.com) — no credit card required.