Browse a library of AI voices, preview them in your content context, compare two at once, or describe the voice you want and let Brandstaq match it. One voice setting applies everywhere — social audio, podcast clips, video narration.
Brandstaq's voice management system gives you granular control over the TTS voices used across all your brand's AI-generated audio content. It's not a single global setting — you can set a default voice for your brand, then override it per content domain (social, content, community) or per deployed agent.
The voice resolution chain works in priority order: agent-level voice override → domain voice from your brand's voice map → brand default voice → admin default. This means your Social Manager agent can have a warmer, more casual voice for TikTok and Instagram content, while your Content Writer uses a more authoritative voice for podcast episode narration and explainer content.
Voice preferences are stored in your brand kit's voice map — a JSON palette that maps domains to specific voice IDs. Change a domain voice once and every agent operating in that domain picks it up on their next run.
Brandstaq provides four ways to select voices through the chat interface:
Browse and preview: Ask your agent to show available voices. Each voice in the catalog shows its provider (Edge TTS for free tiers, OpenAI TTS, ElevenLabs for premium), its characteristics (gender, accent, tone), and a sample preview audio you can play directly in chat.
Side-by-side comparison: Request a comparison of two specific voices. Both generate a preview using the same script — a sentence from your actual content — so you hear the difference in context, not just from a generic sample.
Description matching: Describe what you want. 'A warm, confident voice that sounds like a senior professional, American accent, not too fast.' Brandstaq's matching system scores the entire voice catalog against your description and previews the top three matches for you to choose from.
Direct set: If you already know the voice ID from a previous session or from the catalog, set it directly with a single command.
Every voice preview and generated audio file is stored in your Brandstaq content library with a structured path and 24-hour preview TTL for temporary previews. Permanent audio content — podcast episode narration, video scripts recorded in your brand voice, audio advertisements — gets stored with full retention and appears in your content library alongside video, image, and text content.
Brandstaq's content library uses a plugin-based viewer system. The audio plugin provides a playback interface with progress bar, accessibility controls, and skeleton loading for large files. Audio content from your brand agents and from manual uploads appears in the same library, searchable and filterable by content type, date, and agent.
Voice preview audio cleans up automatically via Celery Beat — the storage cleanup job runs every hour and removes preview files older than 24 hours, keeping your R2 storage costs predictable.
Brandstaq's current voice management system works with AI TTS voices from Edge TTS, OpenAI, and ElevenLabs. Custom voice cloning (uploading recordings to create a voice model that sounds like you) is not currently supported but is on the roadmap. For now, the best approach is using the voice matching feature to find an AI voice that closely matches your natural speaking style.
Voice settings apply to any content generated through Brandstaq's voice pipeline — narrated scripts, audio-first content, and video scripts that include TTS narration. For video content where you record your own footage, voice settings don't apply — those are entirely your recordings. Brandstaq's video pipeline can add AI narration to B-roll or explainer video structures if configured in your pipeline.
Edge TTS voices are included in all plan tiers at no additional token cost. OpenAI TTS and ElevenLabs voices consume tokens based on the character count of the text generated. The cost per operation is set by the platform admin and visible in your token usage dashboard. ElevenLabs voices in particular are priced higher due to their API costs, but produce significantly more natural-sounding output.