Platform reference · Cisco Webex · Meetings · Events · Training · Vidcast · FedRAMP / HITRUST

Webex captions: Webex Meetings, Webex Events, Webex Training, FedRAMP / HITRUST tenant captioning

Cisco Webex anchors at a different segment of the conferencing market than Zoom, Microsoft Teams, or Google Meet. Webex's largest tenants are large-enterprise (typically Fortune 1000 with deep Cisco network infrastructure), federal-government (Webex for Government runs on a separate FedRAMP-authorised stack), healthcare (Webex offers HITRUST and HIPAA-aligned posture), defense and intelligence (ITAR / IL2 / IL4 / IL5 tenancy on Webex for Defense), and large multinational regulated-industry — pharma, financial services, energy, telecom. Captioning a Webex recording catalogue at any of these tenants is a regulated workflow: the captioning vendor has to clear the same FedRAMP / HITRUST / ITAR / EU-data-residency posture as the underlying Webex tenant before it can read a single recording. The captioning surfaces themselves run across Webex Meetings, Webex Events, Webex Training (legacy, still in many tenants), Webex Webinars (modern), Webex Vidcast (async-video), and the Cisco Spark / Webex Teams legacy surface where it persists.

TL;DR

A Webex captioning workflow has five surfaces. (1) Live in-meeting closed captions — Webex's automated speech-to-text closed captioning, with optional CART captioner promotion and third-party live-captioning integration. (2) Webex Assistant — Cisco's in-meeting voice assistant produces a real-time transcript, accessible during the meeting, persisted to the recording. (3) Recording captions — Webex Meetings recordings carry the Assistant transcript plus a post-recording caption-replacement workflow (upload SRT/VTT). (4) Vidcast — Webex's async-video product, with auto-transcript and edit-and-replace. (5) Webex Events session recordings — full-day virtual-event session recordings inheriting the Meetings captioning pipeline. The Webex tenant's FedRAMP / HITRUST / ITAR / EU-data-residency posture defines what external captioning vendors are eligible to touch the recordings; the captioning workflow has to live within that posture.

Webex tenant types — and why they shape the captioning workflow

Webex captions are one workflow shaped by the tenant policy more than by the captioning workflow itself. The relevant tenant types and their constraints:

Webex (commercial). The standard SaaS tenant. Hosted on Cisco's commercial Webex cloud, with EU and APAC data-region options for GDPR / data-residency-sensitive customers. Captioning workflow is the most permissive — third-party captioning vendors with standard SOC 2 + ISO 27001 + DPA posture can typically be brought in.
Webex for Government. Hosted on a FedRAMP-authorised infrastructure stack (FedRAMP Moderate or higher), separated from the commercial cloud. Tenant is restricted to US federal civilian, state, local, and tribal government plus federal-contractor partners. Captioning vendor must meet the FedRAMP authorisation for the data being processed; commercial-only captioning vendors are typically not eligible.
Webex for Defense / Webex IL2 / Webex IL4 / Webex IL5. Hosted on DoD Impact Level-aligned infrastructure. Captioning vendors must hold the corresponding DoD authorisation; very few external vendors do.
Webex Healthcare. Tenant configured for HIPAA-aligned posture, with BAA available. Used by hospital systems, health plans, life-sciences companies. Captioning vendor must be capable of executing a BAA and operating under HIPAA technical safeguards.
Webex EU data residency. EU-resident customers who require GDPR-strict data residency can run on Webex's EU data region; some institutional tenants additionally require that no recording data leaves the EU. Captioning vendors must process within the EU region; non-EU-resident vendors are typically not eligible.
Webex Sovereign Cloud. Country-specific sovereign-cloud Webex tenants (multiple countries) with national-level data-residency requirements.

The captioning workflow does not change much across these tenant types — the same Meetings / Events / Vidcast / Webinars surfaces apply — but the eligible vendor set shrinks materially as the tenant's regulatory posture tightens. For institutions running a mix of tenants (commercial Webex for general meetings, Webex for Government for federal-program-related meetings), separate captioning vendor engagements may be required.

Surface 1 — Live in-meeting closed captions

Webex Meetings supports three live-captioning paths during an active meeting:

Automated captions. Webex's built-in speech-to-text generates real-time captions during the meeting, displayed in the participant's CC overlay. Substantive accuracy lands in the 80–90% band on conversational audio, lower on technical content with named entities, lower again on heavy accents and multi-speaker conversations. Like every other generic auto-caption surface, it produces predictable proper-noun mangling.
Multilingual translation. Webex's auto-captions support real-time translation into a substantial language list. Translation accuracy compounds on top of the source-language transcription accuracy, so proper-noun mangling propagates into the translated cue.
CART captioner. The host promotes a participant to the captioning role and that participant types live captions. This is the supported pathway for institutional CART (Communication Access Realtime Translation) services. Critical for accommodation obligations on US federal-tenant meetings under Section 504 and Section 508.
Third-party live-captioning integration. Webex supports an SDK pattern where a third-party live-captioning service can post captions through a designated API endpoint. Used at large institutions running dedicated CART vendors at scale.

Live captioning is the accommodation surface; it is operationally distinct from the recording captioning surface. Auto-captions are a baseline; manual or third-party CART captioning is the defensible accommodation pathway when an institution has a documented obligation.

Surface 2 — Webex Assistant transcript

Webex Assistant is Cisco's voice-assistant-and-transcription layer that runs during a meeting. It produces a real-time transcript visible in the meeting UI and persists to the recording. Behaviour relevant to captioning:

Per-speaker labelled transcript. The Assistant transcript identifies speakers when speaker-detection is clean. Speaker labels persist into the recording's transcript file.
Highlights and action items. Webex Assistant generates a summary of the meeting plus action items extracted from the transcript. Both inherit the substantive accuracy of the underlying transcript — a mangled product name in the transcript becomes a mangled action item.
Post-meeting transcript editor. The transcript is editable in the Webex web portal; per-cue inline editing or wholesale download-edit-upload-replace.
VTT export. The transcript exports as WebVTT. VTT is the format most LMS, video-host, and downstream captioning workflows accept.
Account-admin enable. Webex Assistant is per-account and per-meeting opt-in in many tenants. Federal-tenant and EU-data-residency tenants frequently disable Assistant by default.

The Assistant's substantive accuracy is the same problem as every generic STT system on technical content. The proper nouns the transcript gets wrong are the proper nouns that distinguish the meeting from generic conversation. Glossary-biased captioning during the post-meeting re-captioning pass is what changes the proper-noun outcome.

Surface 3 — Recording captions

Webex Meetings recordings store the Assistant transcript when Assistant ran, plus a recording-level CC track. The post-recording captioning workflow:

Open the recording in the Webex web portal under "Recordings".
The transcript is exposed as a per-cue list in the playback UI and as a downloadable VTT.
Edit per cue inline OR download the VTT, edit externally, and re-upload as the replacement caption track.
The replacement caption track replaces the Assistant transcript on the recording.
The recording's CC button uses the replacement track.

For institutional retrofit, the API pattern is the production-grade automation: webhook on recording-completed → external captioning service downloads the recording (or pulls the Assistant transcript as a starting draft) → produces glossary-biased VTT → uploads back through the caption-replacement endpoint.

Surface 4 — Vidcast (async-video)

Cisco Webex Vidcast is the async-video product layered into the Webex Suite. Recorders capture screen + camera + audio, the recording uploads to the Webex tenant, and the clip is shareable via a link or embed. The captioning surface mirrors Cloud Recording's:

Auto-transcript on every clip. Same STT pipeline as Meetings; same 80–90% accuracy band on technical content.
Edit-and-replace. Per-cue inline editing or wholesale upload-replace.
VTT export. Standard WebVTT export and import.
Single-speaker default. Most Vidcast clips are single-speaker (the recorder narrating). Auto-transcript accuracy is at the higher end of the band, but proper-noun mangling remains.
Higher publish-as-training fraction. Most Vidcast clips are intentionally produced as training video. SC 1.2.2 substantive accuracy applies from the start.

Vidcast competes with Loom and Microsoft Stream in the async-video category. The captioning failure mode is identical across all three; Vidcast inherits the Cisco enterprise-tenant posture (FedRAMP, HITRUST, EU data residency) that the other two don't always provide.

Surface 5 — Webex Events / Webex Webinars

Webex Events (the modern virtual-event platform) and Webex Webinars (the broadcast-style product) inherit the Meetings captioning pipeline:

Live captions during the session. Auto-captions or CART captioner integration.
Session recordings. Each session recording inherits the Assistant transcript and the caption-replacement workflow.
On-demand consumption. Registered attendees view session recordings asynchronously through the event landing page.
Continuing-education credit obligations. Many Webex Events / Webinars are accredited for continuing-education credits. Accreditation bodies require accessible content; substantive caption accuracy is part of that bar.
Webex Training (legacy). The legacy Webex Training product, still in use at some institutional tenants for instructor-led training, has its own session-recording surface that follows the Meetings caption pipeline. Cisco has been migrating Training tenants to Events / Webinars for several years.

The cross-session asset register for accessibility evidence has to track every session recording. A multi-day Webex Event produces hundreds of hours; institutional retrofit is a substantial captioning project.

Compliance regimes — which apply to Webex tenants

The compliance regimes that bind Webex tenants vary by sector but typically stack:

Section 508. US federal-agency captioning. Federal-tenant Webex Meetings recordings used as official records or made publicly available must clear 36 CFR § 1194 / WCAG 2.0 AA captioning. The Section-508 scope expands when the agency embeds video in public-facing pages or distributes to external audiences.
ADA Title II. State and local-government Webex tenants (state agencies, public universities, public schools) bound to WCAG 2.1 AA on web content and mobile apps post-2026-04-24.
Section 504. Any institution receiving federal financial assistance — universities, hospitals, school districts, non-profit grant recipients — bound to programmatic accessibility. Captions on instructional and program-relevant video.
Section 1557. Healthcare entities under the HHS rule (2024 final rule, 89 Fed. Reg. 37,522) bound to effective communication and WCAG 2.1 AA on web-and-mobile content. Includes patient-education video, telehealth-recording surfaces, and workforce training for federal-program participation.
HIPAA. Webex Healthcare tenants under BAA. HIPAA training captions reference covers the workforce-training mandate at 45 CFR § 164.530(b).
EAA. EU operations bound to the European Accessibility Act (Directive (EU) 2019/882) since 2025-06-28.
EN 301 549. EU public-procurement of ICT including video-conferencing services bound to the harmonised standard. Cisco publishes Webex VPAT / ACR documentation aligned to EN 301 549 for procurement.
AODA. Ontario tenants on the IASR three-year compliance reporting cycle (next major filing window 2026).
CVAA. The FCC's 21st Century Communications and Video Accessibility Act reaches IP-distributed video, distinct from the ADA regime; some Webex content (consumer-facing, IP-distributed) sits in CVAA scope.
FERPA. Education-tenant captioning that includes student-identifiable information — including caption files that name students — is FERPA-regulated. The captioning vendor's data handling must respect FERPA.
SOX, GLBA, FFIEC. Financial-services Webex tenants have additional content-handling requirements that affect captioning vendor selection.

The compliance stack is the reason Webex captioning vendor selection takes longer than Zoom or Loom captioning vendor selection. The captioning vendor has to clear every applicable regime before an InfoSec / privacy / compliance review will sign off on the engagement.

The OCR / DOJ / EU enforcement sampling pattern, applied to a Webex tenant

For a federal-agency Webex tenant or a Webex for Government tenant, the enforcement pattern when an accessibility complaint lands typically follows:

Identify the program. The complaint typically names a specific recorded meeting, training session, or virtual event.
Pull the recording. The agency's compliance officer (or the institution's accessibility coordinator) pulls the Webex Cloud Recording or Vidcast clip from the institutional drive.
Review the captions. Two to three minutes is enough to assess substantive accuracy on the named-entity surface — the regulatory citations, the federal-program acronyms, the technical product terms, the sub-agency or division names.
Read against the audio. Mangled federal-program acronyms (NIH, OCR, OPM, DCAA, DLA, DCSA, DISA, GSA) and regulatory citations (42 USC § 1395, 45 CFR § 164.530(b), 28 CFR Part 35) are the failure pattern that gets flagged in writing.
Sample the back-catalogue. If the named recording fails, the investigator typically samples a half-dozen other recordings to check for a pattern. A pattern triggers a programme-wide finding.
Document the institutional accessibility policy. The accessibility policy, the training records, and the captioning workflow documentation are all part of the response packet.

For a Webex for Defense or DoD IL5 tenant, the equivalent is the DoD IG / Cyber-IG audit pattern — the captioning quality is a smaller component of a larger compliance audit, but it surfaces in the same general way.

Proper-noun failure modes in Webex content

Webex tenant content has a different proper-noun density than commercial Zoom or Loom content. The categories that fail most consistently:

Federal-program acronyms. The full register of agency names (CMS, OPM, OCR, GSA, DLA, DCSA, DISA, DCAA, NIH, NSF, NOAA, NRC, FERC, FCC, FAA, FDA, USDA, EPA, HUD, DOL, Treasury, State, DOJ, HHS, DOC, DOI, DOE, DOT, DHS, OMB) plus sub-agency names plus office-of-this-or-that names.
Regulatory citations. Title / Section / Part / Sub-part conventions across CFR, USC, federal acquisition regulation (FAR, DFARS), NIST publications (SP 800-53, SP 800-171), and the accreditation-standards alphabet (NQF, AHRQ, CDC, etc.).
Military training-command and rank registers. Covered in the Lectora captions reference at depth — TRADOC / FORSCOM / NETC / SOCOM / INDOPACOM / EUCOM / AFRICOM / NORTHCOM / SOUTHCOM / CENTCOM / STRATCOM / SPACECOM / TRANSCOM, E-1 through E-9, O-1 through O-10, plus weapons-systems names.
Intelligence-community vocabulary. ODNI / NSA / CIA / DIA / NGA / NRO and the controlled-information markings (Confidential / Secret / Top Secret / SCI / SAP / FOUO / CUI). Note: ITAR and ITAR-equivalent content should not be sent to commercial captioning vendors at all; the entire workflow must run inside the regulated tenant's vendor ecosystem.
Healthcare drug formulary, procedure names, anatomy. Detailed in medical training captions and Relias captions; same vocabulary applies in Webex Healthcare tenant content.
Cisco product names. The Cisco product catalogue itself is a high-density proper-noun surface in Cisco-internal training content (Catalyst, Nexus, ACI, ASA, Firepower, ISE, Webex itself, IOS XR, IOS XE, NX-OS, etc.). The catalogue is large enough that it's a meaningful glossary input for any Cisco-internal training video.
Financial-services regulatory citations. FINRA / SEC / OCC / FDIC / OFAC / CFPB at depth in financial-services Webex tenants. Detailed in the iSpring captions reference.
Legal-content vocabulary. Latin terms, statutory citations, case names with non-Anglophone parties at depth in legal-services Webex tenants.

The compounding-accuracy property of glossary-biased captioning is most valuable on Webex tenant content because the proper-noun density is highest there and the same vocabulary recurs across thousands of recordings within the same institutional tenant.

The Webex recordings retrofit pattern

For an institutional Webex tenant sitting on years of recordings, the retrofit runs in five phases — same shape as the Zoom retrofit but with a tighter privacy / vendor-eligibility gate at the start:

Inventory. Use the Webex REST API to enumerate every recording (Meetings, Events, Webinars, Vidcast). The Webex APIs expose listing endpoints across all surfaces. Most institutional tenants discover that 30–60% of their recordings have been linked from at least one downstream training surface (LMS course, customer-academy page, internal wiki, agency-training portal); those are the "promoted to training" set.
Vendor-eligibility gate. Before any external captioning vendor touches a recording, the tenant's compliance posture must be matched. FedRAMP authorisation, HITRUST certification, EU-data-residency, ITAR — whichever apply. This step often takes longer than the captioning work itself; the vendor-selection cycle on a Webex for Government tenant typically runs 6–12 weeks.
Triage. Rank by exposure: recordings that are still actively referenced first, regulated-program-related recordings high, public-facing recordings urgent. Recordings nobody links to anymore can be archived rather than re-captioned. The triage cut typically removes 20–40% of the catalogue from retrofit scope.
Re-caption. Replace mangled or absent transcripts with glossary-biased output. The institutional glossary is built once — federal-program acronyms, regulatory citations, sub-agency names, drug formulary if a healthcare tenant, military training-command register if a defense tenant, Cisco product catalogue if a Cisco-internal tenant — and applies to every retrofit asset.
Publish. Push captions back to the originating surface. Replace transcript wholesale on Webex Meetings / Vidcast / Events through the web UI or API. If the recording has been syndicated downstream (LMS, video host, agency portal), push the caption file to that surface as well.
Log. Maintain an asset register: recording URL, originating Webex recording ID, caption file, caption source, reviewer, review date, glossary version, downstream syndication targets. The asset register is the artefact that answers OCR / DOJ / DoD IG / EU enforcement document requests.

See pricing

Where glossary-biased captioning changes the math

The standard institutional Webex retrofit cost calculus pits hand-corrected auto-transcripts against vendor-supplied human captioning. Hand-correction at one to two hours per recording, multiplied by an active institutional Webex catalogue (often 3,000–10,000 hours at a large enterprise or federal-tenant org), multiplied by a $40-per-hour staff or contractor rate, produces a multi-six-figure project. Human captioning at $1.25–$3.00 per minute of video, multiplied by an average 60-minute meeting-recording across that catalogue, produces a similar or larger project.

Glossary-biased captioning collapses both. The institution builds the glossary once. Each minute of video costs a fraction of human-vendor pricing. The accuracy is high enough on the proper-noun surface — federal-program acronyms, regulatory citations, drug formulary, Cisco product names — that the human-review pass collapses from full correction to a quick scrub of the amber-highlighted glossary surface. For a 5,000-hour Webex recording catalogue retrofitted over a six-month window, the GlossCap math (Org plan, sustained throughput) lands well under both alternative paths. See the vendor pricing breakdown.

The high-leverage steady-state pattern is the same as for Zoom: webhook on Webex recording-completed → glossary-biased re-captioning → caption replacement on the recording → asset-register entry. The Webex Assistant chapters, summary, and action items inherit the better transcript, so the downstream signal quality compounds.

FedRAMP, HITRUST, BAA, DPA — the captioning-vendor checklist

Before any external captioning vendor touches a Webex tenant's recordings, the institutional compliance gating typically requires the following documents from the vendor:

SOC 2 Type II. Baseline. Most enterprise InfoSec teams will not engage a captioning vendor without a current SOC 2 Type II report.
ISO 27001 and ISO 27701. Particularly for EU and multinational tenants.
FedRAMP authorisation. For Webex for Government tenants, the captioning vendor must hold a FedRAMP Authorization to Operate (ATO) at the appropriate impact level. FedRAMP Moderate is the typical baseline; FedRAMP High for certain agencies.
HITRUST CSF certification. For Webex Healthcare tenants under HIPAA. HITRUST is the de facto health-sector security framework.
BAA. For HIPAA-applicable content. Captioning vendor must execute a BAA before processing any recording that may include PHI.
DPA under GDPR Article 28. For EU and EU-resident-data tenants. The vendor's sub-processor list must be reviewed and approved.
StateRAMP / regional government authorisations. For state and local government Webex tenants in jurisdictions that require StateRAMP, TX-RAMP, AZ-RAMP, or equivalent.
ITAR / EAR exclusion. ITAR-controlled and certain EAR-controlled content cannot be processed by commercial captioning vendors at all. The institutional compliance officer's sign-off explicitly excludes those categories from the captioning scope.
FERPA-aware data handling. Education-tenant captioning workflows must respect FERPA.
InfoSec questionnaire (CAIQ, SIG, custom). Most Fortune-500 institutional tenants have a specific InfoSec questionnaire; the captioning vendor's prepared answers shorten the engagement cycle.

The captioning-vendor checklist is typically the slowest part of a Webex captioning engagement. The 6–12-week vendor-selection cycle on a federal or large-enterprise Webex tenant is mostly this checklist.

FAQ — Webex captions

Does Webex's auto-transcript clear ADA Title II SC 1.2.2 / Section 508 / EAA / Section 1557?

Webex auto-transcript lands in the same 80–90% substantive-accuracy band as every other generic STT system on technical content. The substantive-accuracy bar SC 1.2.2 enforces is "captions that accurately convey the audio." For technical, regulated, federal-program, healthcare, or any other proper-noun-dense content, auto-transcript virtually always requires correction. The defensible posture is to treat auto-transcript as a draft and run a glossary-biased correction pass before the recording is exposed as official content or training material.

What format does Webex export the transcript as?

WebVTT. The Webex web portal lets you download the Assistant transcript as a VTT file. WebVTT is a superset of SRT — most LMS and video-host upload paths accept VTT directly.

Can I run an external captioning service against Webex for Government recordings?

Only if the captioning vendor holds the matching FedRAMP authorisation for the data being processed (typically FedRAMP Moderate, sometimes FedRAMP High). Most commercial captioning vendors are not FedRAMP-authorised. The institutional compliance officer's eligible-vendor list is much shorter for Webex for Government than for commercial Webex tenants.

What about Webex Healthcare tenants — does the captioning vendor need a BAA?

If the recordings can include PHI (which they often do — internal clinical training conversations referring to specific patient cases), yes. Captioning vendor must execute a BAA before processing. Most captioning vendors that focus on enterprise tenants offer a BAA; vendors focused on commercial / consumer tenants typically don't.

How does Webex Vidcast compare to Zoom Clips and Loom for captioning?

The captioning surfaces are similar — auto-transcript on every clip, edit-and-replace, VTT export. Vidcast inherits Cisco's enterprise-tenant posture (FedRAMP, HITRUST, EU data residency); Loom and Zoom Clips don't always provide the same posture. For tenants that require the regulated posture, Vidcast is the eligible async-video host; for tenants that don't, all three are operationally similar.

What about Webex Training (legacy)?

Webex Training is the legacy instructor-led-training product, still in use at some tenants. The session-recording captioning surface follows the Meetings caption pipeline. Cisco has been migrating Training tenants to Events / Webinars over multiple years; the captioning workflow is identical post-migration.

How does this relate to Cisco's published Webex VPAT / ACR documentation?

Cisco publishes Webex Voluntary Product Accessibility Templates (VPATs) — now Accessibility Conformance Reports (ACRs) under the ITI VPAT 2.5 template — aligned to Section 508, EN 301 549, and WCAG 2.1. The VPAT is the procurement-evidence artefact for institutional tenant selection. The substantive caption-quality bar in any specific institutional context is the institution's responsibility; the VPAT establishes that the platform's surfaces support captioning, not that any specific recording is substantively captioned.

What does the OCR / DoD IG investigation packet typically request for a Webex tenant?

For a video-accessibility complaint, the request typically includes: the recording URL, the caption file attached to it, the institutional accessibility policy, the staff and faculty training records around accessibility, the accommodation-services request log relevant to the complainant, and (for federal tenants) the Section 508 conformance documentation. The asset register described in the retrofit pattern is the artefact that answers the documentation half of the request quickly.