Fresh out of the oven
Spot a bug? These tools are new, so the odd thing may still wobble. If something looks wrong, it probably is. Tell us and we will sort it..
Fresh out of the oven
Spot a bug? These tools are new, so the odd thing may still wobble. If something looks wrong, it probably is. Tell us and we will sort it..
Fresh out of the oven
Spot a bug? These tools are new, so the odd thing may still wobble. If something looks wrong, it probably is. Tell us and we will sort it..
Pinnora's AI podcast generator turns one question into a finished audio episode. Set how many hosts and how long, and it plans the segments, scripts each one as speaker turns, and voices them separately. What lands is an MP3 plus a speaker transcript where every turn carries its start time, so clicking a line seeks the audio.
The tool is open above. This is the short version of what to do with it.
40 to 80 words: the question the hosts disagree about, and where each of them stands. A topic list gives you a flat read.
Length goes up to 60 minutes and drives the structure, since the outline plans about one segment per three minutes.
Hosts runs 1 to 4, Vibe sets the register across five options, and Brand weave decides how often your product comes up at all.
The episode lands with the player on the left and a speaker transcript on the right. Click any line to jump the audio, then download the MP3.
Short answers to what people ask before they start.
Hosts runs 1 to 4 and starts at 2. Each host is voiced separately, so a two-host episode is two distinct voices trading turns rather than one narrator reading both sides. In the full workspace you go further: name each host, give them a persona (founder, skeptic, analyst, enthusiast, comedian, industry vet), and pick their voice. It refuses to generate while two hosts share one voice.
Length goes up to 60 minutes and starts at 10. The script is planned to that number rather than trimmed to it afterwards. The outline pass turns your target into segments, roughly one per three minutes and between 4 and 12 in total, and each body segment is written to carry about three minutes of dialogue. A cold open, some intro banter and a close bracket the body.
It pulls your project's brand foundation, research, strategy, winning creatives and stored brand memory into the scripting prompts. None of that is read aloud. It shows up as vocabulary, the customer problems the hosts name, and the proof they reach for, so the episode sounds like people who work in your category. Grounding soft-fails, so a bare project still renders, with less specificity.
One MP3 per run, plus the full transcript. In the viewer the player sits on the left and the transcript on the right, speaker by speaker. Every turn carries its start time, so clicking a line seeks the audio to it. The download buttons hand you episode.mp3 and transcript.md, and the episode also lands in the project Library.
Audio is billed by the character of speech synthesized, so cost tracks how long the episode runs and how much the hosts actually say. A 10 minute two-host episode costs a fraction of a 60 minute one. When the episode needs a separate mix pass, that pass adds a small compute charge. Charges are keyed per run and per turn, so a retry never bills the same speech twice.
Yes, from the full workspace. Intro jingle and Outro jingle are toggles that stitch your project's most recent jingle to the front and back of the episode. Make the jingle first with the jingle tool, then run the podcast. With no jingle in the project the toggles find nothing to stitch, and the episode ships without bumpers.
The rest of the family. Every one is free to try and runs on the same engine.
