Real questions.
Honest answers.

flexVox is an iOS app that turns multi-speaker scripts into produced podcast audio. You paste a dialogue script, assign AI voices to each character, generate speech and sound effects, then mix and export the result — all on iPhone or iPad.

For real audio generation, yes. You'll need an ElevenLabs API key (free tier available at elevenlabs.io). flexVox stores the key in the iOS Keychain. A built-in demo mode works without any account — it generates silent placeholder audio so you can explore every screen and feature first.

iPhone and iPad running iOS 26.4 or later.

Script import, parsing, and review work offline. Audio generation requires an internet connection to reach the ElevenLabs API. Demo mode works fully offline.

Colon format (HOST: Welcome), bracket format ([Host] Welcome), parenthesis format ((Host) Welcome), and a standalone name on its own line followed by dialogue beneath. SFX and music tags use [SFX: prompt (duration)] and [Music: prompt (duration)], with optional modifiers @underlay, volume=, loop, and influence=.

Every attribution has a confidence score. Low-confidence turns are highlighted for review. Tap any turn to reassign, batch-assign unreviewed turns, or merge duplicate speakers.

Yes. The Script Edit sheet lets you modify the raw text and re-parse — but re-parsing replaces all existing speaker assignments and deletes any generated audio for the project. flexVox warns before doing this.

No hard limit on the number of speakers. Each gets their own voice profile. Dialogue batch mode batches up to 10 unique voices per API call; larger casts auto-split across batches.

Yes. Each voice profile has independent sliders for stability, similarity boost, style exaggeration, and speed, plus a speaker boost toggle. v3 models use a simplified control set.

ElevenLabs v3 audio directives that control how a voice performs a line — like [happy], [whispers], [laughs], [sighs]. Place them inline in dialogue text. Stack tags for complex emotions ([angry][laughing]). Write custom descriptive tags like [trying to sound brave] — v3 interprets natural language in brackets.

Yes. Use Enhance Expression Tags from the toolbar (or long-press a single turn). The app uses Apple Intelligence on-device when available, or falls back to your configured AI provider (OpenAI or Claude). Suggestions are context-aware — surrounding dialogue informs each tag. Review each before applying.

@underlay plays the audio under dialogue instead of sequentially. volume=0.3 sets underlay volume (0–1). loop generates seamless looping audio. influence=0.8 controls how literally the AI interprets the prompt (0–1). Example: [SFX: rain on windows (5s) @underlay volume=0.3 loop].

Audio that was already generated is preserved. Resume picks up where it stopped, skipping completed turns. Network errors and rate limits are retried automatically with exponential backoff.

Yes. Swipe the turn or use the context menu to regenerate. The new audio saves as a variant — your previous take is not overwritten.

Multiple takes per turn. When you regenerate, the new take is added alongside the existing ones. Play each, pick the best, delete the rest.

Two levels: a project-wide default pause set in post-production, and per-turn custom pauses set with the Adjust Pauses toggle and individual sliders (0–10 seconds). Custom pauses override the default for that turn.

M4A (AAC). Normalized to your chosen export preset's LUFS target.

A container for a series of related episodes. It stores a persistent cast bible with character profiles and voice assignments, a default format and tone, audio identity (intro / outro music), and an episode template. New episodes inherit all of it.

No. Shows are entirely optional. Standalone projects work exactly as they did before. You can promote any project to a series later by long-pressing it and choosing "Create Show from This."

Cast members are stored per production. To use the same character in a different show, add them again in that show's cast bible. Voice assignments (ElevenLabs voice IDs) are portable since they reference the same voice catalog.

OpenAI (GPT-4o) and Anthropic (Claude). Set up your API key in Settings under AI Script Writing. Switch between providers at any time.

Yes. The AI Prompt Sheet has Copy to Clipboard. Paste it into ChatGPT, Claude, or any other tool, then paste the generated script back into flexVox.

The app shows an estimated cost before generation based on your prompt length and selected provider. Costs depend on your provider's pricing. flexVox itself charges nothing for this feature — you pay your AI provider directly.

In the iOS Keychain — the same secure storage the operating system uses for passwords. It is never written to a plain file or sent anywhere other than the ElevenLabs API.

No analytics frameworks, tracking SDKs, or server-side data collection. Your scripts, audio files, and projects stay on your device. The only network calls are to ElevenLabs (audio) and your AI provider of choice (script writing).