Skip to main content

Chat with safeguard-class models, minimal retention mindset, and downloadable audio

How we approach AI chat, provider choices, and keeping voice data under your control.

Protected and vetted models

The chat area connects to curated large-language models — including options designed with stronger safety and alignment characteristics (sometimes referred to as “safeguard” or enterprise-friendly stacks). Administrators can enable or disable providers and models per deployment so that only reviewed endpoints are exposed.

Where we route through specialized high-speed hosts (for example low-latency inference clouds), we select configurations that fit our security and compliance posture. Model lists and pricing in credits are visible inside the app for transparency.

Retention and your content

WebVoice is built so that routine chat turns are not used to train your private tenant model — we do not operate a public “learn from all users” chat product. Provider APIs may apply their own short technical retention for abuse prevention; we configure flows to minimize unnecessary storage and point you to our Privacy and AI Policy for details.

For voice notes and TTS/STT, you decide what to keep in your library; exports and deletions are governed by the same account tools and legal terms.

Audio with MP3 you can carry away

When you synthesize speech or record voice memos, the platform favours efficient MP3 assets you can download, archive offline, or attach to tickets. That matters for podcasts, customer support, and regulated environments where a portable file is part of the record.

If you use integrations or the API, you can still land the same bytes on your servers and apply your organisation’s retention and backup rules there.

Further reading

For a technical comparison of local vs cloud models and security notes, see our models and security pages linked from the in-app Info section when logged in. The public AI Policy summarises ethical use and limitations.

Summary

  • Chat uses administrator-approved models and providers
  • No “train on your chats” consumer loop — see policies for provider retention
  • Voice outputs available as downloadable MP3 where the feature applies
  • Combine with API for server-side pipelines under your policies

Trusted for production voice workloads

Inside the product — app screenshots, workflow, and reserved logo slots on the main site.

Product teams, agencies, and developers use WebVoice for TTS, STT, chat, and API-first integrations — from prototypes to customer-facing apps.

SaaS & product
Agencies
Education
Internal tools
50K+

Hours of audio synthesized & transcribed monthly (illustrative range)

30+

Neural voices and locales in the catalogue (varies by deployment)

REST

Same credit wallet for browser app and documented HTTP API

Figures are indicative and depend on traffic and configuration.

“We shipped read-aloud and STT in one sprint — the API matched what we tested in the UI.”

“Credits per feature make finance happy — we can forecast TTS vs chat separately.”

“Low-latency Groq routes for chat let us keep UX snappy without a separate vendor.”

Quotes represent typical feedback patterns; not attributed to specific customers.

Frequently asked questions

Administrators curate which providers and model IDs appear. You pick per thread; credits per request are shown before you send.

Where the feature applies, outputs such as MP3 can be downloaded for your records — check the product limits for your account tier.

Routine turns are not used to train your private tenant model. Third-party hosts may keep short-lived logs for abuse detection — see policies for detail.

Safeguard-class options add an extra policy-aligned layer for sensitive use cases; you can still combine them with voice workflows.

The product may prompt for acceptance where required; links are also in the legal footer.

Ready to try WebVoice?

Get Started API documentation