Podcast by roles: how to turn your workflow into a dialogue
The 'Podcast by roles' tool creates a podcast script in dialogue format: you describe an idea, set host roles and duration, and get a script with lines, segments and natural transitions. In this article we break down how to use it for the topic 'Workflow for everyday tasks': what data to feed in, how to formulate roles, how to check the result and where the limits lie. Important: the tool generates text only. It does not record audio, does not create video and does not publish an episode. All you get is plain text that you can copy or download as TXT.
Open tool: Podcast script writer ↗What exactly does the tool do
'Podcast by roles' is a text script generator. Three fields are input: 'Your text or idea', 'Role or characters' and 'Duration or length'. The first field accepts up to 8000 characters, the second up to 1000 characters. Output is one result in a separate panel that can be copied or downloaded as TXT.
The tool does not set style presets, has no sliders and no levels of editing. It is a form with text fields and a Generate button. You choose the output language, press the button and get a dialogue. No automatic adaptation to genre, tempo or speaking style is provided - you describe all that in words in the roles field.
Separately, it is worth noting what the tool does not do: it does not generate audio or video, does not clone voices, does not create photorealistic images. If images are needed, these are vector compositions in SVG and PNG. Speech, if needed separately, is synthesized via eSpeak in WAV. Subtitles are in SRT. No DOCX or PDF export, no batch processing, no auto-save.
Example input data: input - 'podcast about how to get order in daily work tasks'; role - 'host Anya, organized; host Max, skeptic'; length - '5 minutes'.
How to prepare input data
Start with an idea, not a ready-made text. In the 'Your text or idea' field describe the topic, audience and episode goal. For the topic 'Workflow for everyday tasks' a suitable formulation is: 'Conversation for beginning specialists on how to declutter your day and not lose important things'. The more specific the goal, the less water in the result.
In the roles field describe characters by character and function in the dialogue. Not just 'host 1 and host 2', but for example: 'Anya - methodical, loves lists; Max - doubts, asks uncomfortable questions'. Such a pair gives natural conflict of opinions, and therefore - a living dialogue instead of a monologue.
Specify duration realistically. A five-minute episode is about 600-750 words of dialogue. If you request '30 minutes', the model may stretch the text with repetitions. Better to make several short scenarios than one long and rambling one. Remember: the first input field is limited to 8000 characters, and the details field to 1000 characters.
Bad: 'make a podcast'. Good: 'dialogue for 5 minutes, two hosts, topic - morning task review, audience - beginners'.
Script structure: segments and transitions
A good podcast script rests on three pillars: introduction, main body and conclusion. In the introduction hosts define the topic and promise value. In the main body they discuss two or three specific steps. In the finale - a short call to action. The tool gathers this structure automatically if you set it in the idea.
Segments within the main body are conveniently built on the scheme 'problem - technique - example'. For example: problem - tasks pile up; technique - rule of three main tasks; example - how it looks on Monday. Such a scheme gives natural transitions: 'What to do if there are still more than three tasks?' - and the second host answers.
Transitions between segments - the weakest point of automatic scripts. To strengthen them, describe in roles the speaking style: who summarizes, who asks questions, who jokes. Then the lines will not sound like two independent monologues. Check that each transition answers the question 'why are we talking about this now'.
Segment: 'Problem: task list grows faster than you close them. Technique: three main tasks a day. Example: Monday, 9:00, you choose three items and postpone the rest'.
Practical example for the topic about workflow
Suppose you want an episode about everyday tasks. In the idea field write: 'Podcast for those who drown in small things. Goal - show a simple ritual: morning choose three tasks, evening close the tails'. In the roles field: 'Lena - practitioner, speaks short phrases; Igor - analyst, likes to clarify details'. Duration: '6 minutes'.
The tool will return a dialogue where Lena suggests the ritual and Igor checks its strength: 'What if there is only one task, but big?', 'What to do with sudden urgencies?'. These are natural transitions - they arise from characters, not from a template.
Check the result against three criteria: is there an introduction and finale; do the same advices repeat; do the lines sound like speech, not like an article. If the dialogue turns out too smooth, add conflict to roles: one host doubts, the other insists. This will enliven the text.
Result fragment: 'Lena: Morning choose three things. Igor: Only three? What if there are twenty? Lena: Then twenty is not a list, but anxiety. Start with three'.
How to check and refine the result
First read the script aloud. The hosts' speech should sound natural: short phrases, questions, clarifications. If lines are long and bookish, shorten them manually - the tool has no edit mode and no change highlighting, so refinement is done in your text editor.
Check roles balance. In a good dialogue hosts speak roughly equally and react to each other. If one character only asks questions and the other reads a lecture, rewrite some lines. Also make sure segments do not duplicate each other in meaning.
Separately check facts and numbers. The model may err, and service availability varies. Do not rely on generated statistical data: if numbers appear in the script, replace them with your own or remove them. Responsibility for the episode content remains with you.
Checklist: introduction and finale present; roles balanced; no repeats; numbers checked; text reads aloud without stumbles.
Limitations and technical boundaries
The tool works only with text. It does not record audio, does not create video and does not publish an episode. If you need narration, it is done separately: eSpeak speech in WAV, subtitles in SRT. Images - vector, in SVG and PNG, without photorealism. Export - only plain text TXT; DOCX, PDF, batch processing and auto-save are not supported.
There are also quantitative limits: first field up to 8000 characters, details field up to 1000 characters, speech up to 3000 characters, images up to 4 MB, browser audio up to 20 MB or 5 minutes. Registration, premium plan, integrations and email delivery are not provided.
Finally, remember quality: model answers may contain errors, and service availability is not guaranteed. The script is a draft, not a ready episode. It must be read, corrected and only then used for recording. The tool gives no guarantees of hitting recommendations or audience growth.
If you need a 20-minute script, make four linked 5-minute episodes - so the text stays dense.
'Podcast by roles' is a narrow but useful tool for those who need a text framework for a dialogue. It helps quickly assemble structure: roles, segments, transitions. But it does not write the final episode for you and does not replace editing. Set a specific idea, describe the hosts' characters, specify a realistic duration - and check the result aloud. For the topic 'Workflow for everyday tasks' this is enough to get a practical five-to-six-minute script that only needs polishing to your speaking style.
Open tool: Podcast script writer ↗Interface translations are machine-assisted and may need refinement.