Changelog

What changed?

You can now choose between two text to speech modes when producing audio: Good (default) and Best.

• Good: Fast, reliable, natural enough for long scripts at low cost.
• Best: Premium voice generation with richer prosody and emotional nuance.

Why it matters

Creators report common pain points:

• Moments that should feel emotional sounding flat
• Intros and ad reads needing more presence and warmth
• Tone drifting over long episodes
• Mixed language terms or brand names losing correct stress

Good handles these for most use cases. When you need that last bit of realism for high stakes audio, Best is available.

How it works

• Mode is set per episode in the editor.
• Use the Good/Best toggle in the editor header.
• Switching modes removes existing generated audio for that episode and requires regeneration, which uses credits.
• Voices reset to provider defaults after switching. Adjust voices as needed.

When to use each

Use Good for:

• Educational series, explainers, internal training
• Drafts, fast iterations, bulk catalog updates
• Episodes with limited emotional range

Use Best for:

• Trailers, ads, brand intros, launch announcements
• Narrative segments that require tension or empathy
• Final masters for public release

Pricing

• Good uses standard credits.
• Best uses 10x the credits. Choose it only where the extra realism is required.

Getting started

Open any episode, use the Good/Best toggle in the editor, then regenerate lines or select Generate all audio.

Stan

What is new

You can now create podcasts directly in your audience's language. Pick Spanish, Hindi, German, Japanese, Arabic, Swahili (and 60+ more) and the script is written natively in that language. No after‑the‑fact translation feel.

Why it matters

Listening in your own language lowers cognitive load. Complex ideas, technical walkthroughs or nuanced stories land faster and stick longer. This helps you:

• Make expert content accessible to teams in different regions.
• Reach new markets without rewriting everything manually.
• Let learners follow along without juggling a second tab for translation.
• Offer inclusive, language‑first onboarding or education series.

How cloning works

Language for an episode is fixed once created so timing, tone and future audio stay consistent. Need another language? Click Create language version. We duplicate the source inputs and generate a fresh script in the new language while the original stays untouched. Each version progresses independently (draft → refine → voice).

Use cases

• Publish an AI safety explainer simultaneously in English, Portuguese and Korean.
• Turn a long technical architecture brief into localized training series.
• Test demand in a new region with a single cloned episode before committing.
• Repurpose evergreen thought leadership into a multi‑language learning path.

Tell us which language or dialect you want polished voices and style presets for next.

{{EM}} Stan.

What changed?

You can now create podcasts from simple text prompts or ideas. Just describe what you want, and our AI will generate a complete podcast script with natural dialogue between two hosts.

How it works

1. Describe your idea: Enter a few sentences about what you want your podcast to cover
2. Choose your style: Pick from educational, conversational, narrative, or deep dive formats
3. Set duration: Specify how long you want the episode (1-60 minutes)
4. AI generates: Our system creates a full script with natural host dialogue
5. Customize & produce: Edit the script, choose voices, and generate audio

Perfect for

• Quick content creation from brief ideas or concepts
• Educational content explaining complex topics to specific audiences
• Thought leadership pieces on industry trends or insights
• Content repurposing from notes, outlines, or bullet points
• Brainstorming sessions turned into structured discussions

Example prompts

• "A 5-minute podcast explaining machine learning basics for beginners, using simple analogies"
• "Deep dive into the future of remote work, covering tools, culture, and productivity trends"
• "Conversational episode about sustainable investing for millennials, practical and actionable"

Why it matters

This dramatically lowers the barrier to podcast creation. No need for detailed outlines or scripts—just share your idea and let AI handle the structure, flow, and dialogue. Perfect for subject matter experts who want to share knowledge without spending hours on script development.

Pro tips

• Be specific about your target audience (e.g., "for beginners", "for executives")
• Mention the tone you want (professional, casual, energetic)
• Include key points you want covered
• Specify real-world applications or examples to include

— Stan

What changed?

The generator can now securely access the public internet while crafting your podcast script. That means:

• Coverage of breaking news and emerging stories.
• Inclusion of fresh stats, prices, rankings, and market data.
• Automatic link enrichment so sources are transparently cited.
• Smarter context disambiguation (the model double‑checks ambiguous facts).

How it works

For every generation, the system now automatically:

1. Builds focused search queries based on your topic.
2. Fetches and cleans candidate pages (paywall‑safe + robots aware).
3. Extracts structured nuggets (claims, figures, quotes, dates).
4. Runs a fact consistency pass to drop contradictions.
5. Injects a compact, attributed research digest into the model context.

Why it matters

This closes the gap between evergreen knowledge and what's happening right now. Perfect for market wrap‑ups, tech trend briefings, sports recaps, policy updates, or rapid commentary formats.

Safeguards

• Source diversity scoring prevents single‑site echoing.
• Automatic date validation flags stale articles.
• Hallucination guardrails compare claimed numbers to extracted originals.
• Applies only publicly accessible content and respects robots directives.

Getting started

Nothing to enable—live web research is on by default. Optionally provide focus keywords to steer retrieval and framing.

Let us know what real‑time formats you create—this unlocks a whole new style of timely audio.

— Stan