WellSaid Producer Moves AI Voiceover Into MCP-Connected Workflows

WellSaid launched Producer on November 10, 2026, placing its full voiceover production pipeline — casting, direction, pronunciation, review and versioning — inside AI assistants such as Claude, ChatGPT and Copilot through Model Context Protocol connections. The release also generates audio natively at 48kHz, the standard sample rate for film, video and broadcast, which WellSaid calls a first for a voiceover production platform.

For HR and learning teams, the practical question is adoption: can training narration, compliance updates and product videos now be produced and revised inside the tools those teams already use, without losing control of pronunciation, versioning or actor licensing? WellSaid has built Producer around answering yes. The company's MCP server connects its production tools to AI assistants and to customers' own software. WellSaid Studio, the existing web workspace, remains available for hands-on editing, and the two environments share voices, pronunciation libraries and projects.

What does the workflow actually look like?

The documented process starts with a script drawn from an MCP-connected assistant, content platform or project management tool. From there, teams can:

• audition from hundreds of available voices;
• direct pacing and emphasis in plain language;
• send the result for approval;
• push finished audio and captions into a content management system, digital asset manager or learning management system.

Every piece of finished audio carries an embedded rights and license record. Audio created through Producer receives versions and edit tracking in Studio, so the audit trail survives even when production happens outside WellSaid's own interface.

Maintenance is a core part of the pitch. When a script, product name or policy changes, Producer updates every affected voiceover in one pass using the approved voice and direction. Reviewers then assess the new versions instead of rebuilding files one by one. For enterprises running recurring courses or policy videos, that addresses the costly tail of voiceover work — the revisions, not the initial recording.

Chief executive Benjamin Dorr framed the launch as a response to how teams now work: they combine capabilities from different tools, and WellSaid is focusing on accurate delivery, a consistent brand voice and licensing that protects actors.

How does pronunciation control work at scale?

Producer applies what WellSaid calls agentic style guides, which enforce pronunciation choices for brand names, product names, acronyms and technical terms across every connected workflow. The underlying model was trained on more than 500,000 Oxford Languages transcriptions, WellSaid said.

That dataset comes from a collaboration announced on February 6, 2024. Oxford Languages supplied English transcriptions with and without syllable divisions, plus audio for most words, to train WellSaid's Respelling system. Respelling lets users shape the pronunciation of real or invented words, and Producer now carries that control across connected tools. A preferred pronunciation stays consistent as content gets revised.

On fidelity, Producer's native 48kHz output allows full-quality audio to enter a video timeline without upsampling. The capability is separate from the 96kHz output WellSaid has supported since 2025 for commercial editing and mixing, and it sets a higher bar than industrial and interactive voice response applications.

What happens to actor consent and pay?

The same licensing applies whether a team works in Studio or through Producer. WellSaid says every voice licensed for commercial use comes from a professional actor who consented and who is paid when the voice is used. Terms for each voice are visible in both environments, and the license record stays attached to finished audio even when production occurs in an AI assistant or a customer-built tool. Customers can also build libraries of custom digital voice replicas for their brands under the same consent-and-licensing approach.

Alicia Harris, an actor behind one of the platform's voices, said the product makes her voice easier to use while customers gain efficiency and actors "remain part of the process and are compensated." Her statement points to the labor dimension buyers should weigh: automation changes where voiceover is made, but this model keeps consent and compensation attached to use.

Producer also allows voice models from other providers to run alongside WellSaid voices in a single workflow. WellSaid builds its own models, works with other models trained on commercially cleared data, and said it plans to expand Producer's talent and licensing options — a signal that procurement teams should expect the multi-vendor voice question to grow more common, not less.