Skip to content

beatra-ai/beatra-skills / plugins/ai-podcast-voiceover/skills/ai-podcast-voiceover

ai-podcast-voiceover

Verified

Turn an article, notes, or a finished script into a listener-ready solo podcast episode with a consistent host voice. This AI podcast voice generator and AI podcast narration service adapts supplied material into a speakable podcast script, sets names and specialist terms for clear pronunciation, and creates MP3 podcast audio with natural pacing. Use this podcast voiceover AI and text-to-speech podcast service for article-to-podcast audio, news briefings, expert commentary, and knowledge shows, then carry the host direction into the next episode.

AI Podcast Voiceover

Work as an episode producer for one recurring host or a multi-host show: begin with the show profile, turn supplied material or a named topic into a listener-friendly script, prepare expressive pacing with Smart Voice Director, freeze the host cast and model path, prove a new direction with a short audition or sample, then synthesize and record the episode audio, optional music bed, stingers, and loudness finishing.

Scope and routes

Use this workflow for supplied articles, notes, outlines, newsletters, or final scripts, and for a topic alone, when they become a solo podcast episode, a two-host or guest dialogue episode, trailer, opener, closer, or ordered episode parts. Keep recurring host directions, pronunciation glossaries, and delivery facts together when the show will continue.

  • Single-host episodes and narration segments: use beatra.speech.synthesize (with beatra.speech.prepare_script and free beatra.speech.validate).
  • Multi-host or guest conversations: use native beatra.speech.synthesize_dialogue (with free beatra.speech.validate_dialogue) for cohesive scenes with voices whose capabilities.supports_dialogue == true. If a dialogue request is refused as too long, split it at scene or turn boundaries as the refusal message says, passing previous_text and next_text across the boundaries. Fall back to segmented beatra.speech.synthesize per turn when the cast is refused or needs unsupported dialects.

Route a manuscript or chapter-led course narration to ai-audiobook-narration; a standalone short social, ad, or promo read to short-form-voiceover-audio; and supplied approved scripts across several languages to ai-multilingual-dubbing. When a user wants to clone their own podcast host or guest voice for this show, clone and manage that voice directly here (beatra.voices.clone / beatra.voices.activate / beatra.voices.delete); route standalone clone-only work to voice-cloning-studio or beatra-ai-voice-studio. Keep supplied material, episode order, language, and show direction intact when switching routes.

Keep music beds (beatra.music.generate), transition stingers or looping ambient beds (beatra.audio.generate_sfx), and single-file loudness, trim, fade, or vocal pre-cleaning (beatra.audio.process) here as explicitly scoped paid steps — never send those requests away just because audio finishing is involved, and never call any paid finishing step free. Route multi-track mixdown or ducking across tracks, RSS feed hosting, and distribution publishing out to dedicated post-production or podcast-hosting workflows, keeping the user's multi-track requirements intact (offer unmixed ordered dialogue/music/SFX stems only when the user accepts assembling them in their editor).

Open the show

The hard inputs are the show and its audience: who is listening, and what this episode should accomplish. Look first for ./<show>/show-profile.json and reuse that archive when it exists. Reuse stored directions only after each frozen voice_id is re-validated against the current beatra.voices.list and beatra.models.list output. For a new show, establish the intended listener, host cast, language, recurring opener and closer, pronunciation glossary, and delivery conventions in a user-owned working directory. Confirm separately if the user wants intro/outro music (beatra.music.generate), transition stingers or looping ambience (beatra.audio.generate_sfx), or broadcast loudness finishing (beatra.audio.process). Follow the show profile.

Prepare the episode script

Accept an article, notes, an outline, a final script, or a topic alone and adapt it into an episode listeners can follow. Start from the show's intended listener; if supplied material could serve materially different listener goals, ask for the intended takeaway before drafting. A topic alone does not wait on that question: infer the listener and draft in the same turn. Otherwise state the inferred episode focus in the draft for approval. When the user supplies material, every fact, opinion, number, and claim in the spoken script comes from that material or from an edit the user confirmed. An evergreen topic alone is drafted from widely accepted general knowledge, and a time-sensitive topic gets its opening, structure, transitions, and close without facts until the user chooses sources or pastes material. The three-way rule is in episode script guidance. If a supplied link or file cannot be read, stop and ask for the text; never switch the topic or the source. A topic-only request defaults to the bulletin tier, an estimated 3-5 minutes, and the user can choose a longer tier.

Show the episode focus and spoken draft section by section, including the opening, closer, and pronunciation glossary, then wait for approval before paid audio is proposed. If the user wants only a script, deliver the approved script and stop. Use episode script guidance for the material boundary, listener focus, spoken adaptation, pronunciation, and length handling.

Before synthesizing speech, use Smart Voice Director (beatra.speech.prepare_script, {"capability": "speech_script"}) when pacing, pause planning, emotional arc, or pronunciation compilation benefits from AI speech direction rather than guessing pauses or manually authoring braced performance tags (never write braced tags by hand). beatra.speech.prepare_script preserves every spoken word untouched, inspects voice/model capabilities, and returns directed plus scripts[] (inspect warnings if any synthesis is null; pass ready scripts[].synthesis unchanged plus a fresh client_request_id to beatra.speech.synthesize). Keep at most 2 concurrent beatra.speech.prepare_script tasks; within 48 hours for identical text, instructions, language, and context, calling beatra.speech.prepare_script again changing only voice (and model) reuses the earlier direction at zero charge (the balance must still cover admission).

Freeze voices, optional clones, and model paths

Browse presets and existing clones with beatra.voices.list (bilingual search, limit up to 500, cursor, plus filters such as language, gender, age_group, use_case, emotion, model, supports_dialogue, supports_direction, supports_performance_tags, supports_context) and inspect all 11 capabilities.* metadata fields returned for each voice. Call free beatra.voices.preview (preview_url, kind) so the user can audition voices before committing credits. Freeze opaque, ready voice_id handles — never submit a display name or guess from memory. Without a live result, do not name voices, models, prices, or previews.

If the user wants to clone a host or guest voice for this show, follow the three mandatory confirmations in speech, voices, cloning, and audio finishing:

  1. Confirmation #1 — Consent (consent_attested: true): require explicit user confirmation in the current conversation that they own the voice or hold permission to clone it before uploading (scripts/mcp_client.py upload). Access to a file or public recording is never consent.
  2. Sample noise check: if the sample has background music or room/street noise, propose paid beatra.audio.process (operation="isolate_voice", {"capability": "audio_voice_isolation"}, 4.6 s – 5 min, <= 100 MB) on its own production card before cloning.
  3. Clone model choice: select the clone model as described in speech, voices, cloning, and audio finishing §5. When a cloned voice will speak in a two-host or guest dialogue scene, select the clone card that states multi-speaker dialogue; a solo episode keeps the default. Then show the production card with the live clone price; after the user approves it, submit once and handle any refusal per billing, errors, and recovery.
  4. Confirmation #2 — Activation (beatra.voices.activate): when a clone result or a listed voice has status pending_activation, follow §5 Step E of speech, voices, cloning, and audio finishing before using the voice.
  5. Confirmation #3 — Two-step deletion (beatra.voices.delete): to delete a cloned voice later, first call beatra.voices.delete with voice_id only (no confirmation_token), show the returned paid_slot_warning, irreversible_notice, and 15-minute expiry, wait for explicit user approval, then call beatra.voices.delete with voice_id and confirmation_token.

Before quoting models or prices, call beatra.models.list for the exact capability (text_to_speech, text_to_dialogue, speech_script, voice_clone, text_to_music, sound_effects, audio_process, or audio_voice_isolation). An explicit model must be current, compatible with every selected voice_id, and support the target BCP-47 language. For auto in single-speaker synthesis, retain only live models in models.list.auto.candidate_order ∩ voice.compatible_models and use auto only when nonempty and every candidate supports the language. Default to model: "auto", MP3, speed 1.0, volume 1.0, pitch 0, subtitle_enable: true (never pass subtitles), and no explicit sample rate or emotion unless needed. Follow voice, delivery, and recovery and speech, voices, cloning, and audio finishing.

Golden path

Lock the show and audience first. Write the episode for the ear (and attribute each speaker cleanly for multi-host episodes). Direct complex pacing with beatra.speech.prepare_script when helpful, run free beatra.voices.preview and free beatra.speech.validate or beatra.speech.validate_dialogue, then confirm the production card. Prove a new direction with a short sample when applicable, then run episode synthesis or scene-chunked dialogue. Present any beatra.music.generate, beatra.audio.generate_sfx, or paid beatra.audio.process finishing step (normalize_loudness, trim, fade) on separate cards or itemized paid lines — never fold them into a speech approval and never label them free.

Before the connection is ready

When Allow is still pending at the end of a turn, finish and deliver everything that needs no connection, in the format the user asked for: the show profile, the full episode script, speaker turns if multi-host, the pronunciation glossary, three title options, the episode description and show notes, and estimated chapters. All Beatra MCP calls wait until after Allow. Say in one line that Allow is the one-time sign-in that connects their Beatra account, not an approval of any paid card, and that the workflow continues on their next message after it. Do not fill the gap with another tool or service.

Price and confirm

Free tools cost zero credits: beatra.voices.list, beatra.voices.preview, beatra.models.list, beatra.speech.validate, beatra.speech.validate_dialogue, beatra.voices.delete step 1, beatra.tasks.get, beatra.tasks.list, beatra.wallet.get, and beatra.wallet.ledger. Keep beatra.wallet.get and beatra.wallet.ledger as optional read-only lookups when the user asks about balance or charges — never make beatra.wallet.get a mandatory gate before submit, and never ask the user to confirm they have enough credits.

Every call to beatra.speech.prepare_script, beatra.speech.synthesize, beatra.speech.synthesize_dialogue, beatra.voices.clone, beatra.voices.activate, beatra.music.generate, beatra.audio.generate_sfx, and beatra.audio.process is paid (never quote hardcoded prices or call finishing steps free). Before the first paid step, show a clear production card stating all six fields:

  1. Work: exact capability, text/scene/clip scope, weighted characters or clip duration, frozen voice_id(s) and settings, plus any music bed, SFX, or beatra.audio.process operation (trim, fade, normalize_loudness, isolate_voice).
  2. Credits: live quote from beatra.models.list / beatra.speech.validate / beatra.speech.validate_dialogue (never guess; show range or upper bound when auto resolves across multiple cards).
  3. Count: exact number of paid calls approval authorizes (including separate scene dialogue chunks or finishing calls).
  4. Identity: one fresh caller-generated opaque client_request_id per paid call. Do not automatically retry any paid step; changed parameters require a new card and new ID.
  5. If we stop here: name what the customer keeps at zero additional charge (approved script, titles, notes, pronunciation glossary, audition picks, any delivered clips, or plainly nothing paid yet before the first submit).
  6. If the balance is insufficient: relay Beatra's official message and exact top-up URL verbatim without modifying the URL, wait until the user confirms they have topped up, then retry the same frozen client_request_id.

On a first solo episode or after a host/control/language change, begin with a small paid sample card (opener plus hardest passage). Sample approval never authorizes the episode; speech never authorizes music/SFX/finishing; and music/SFX/finishing never authorize speech. If beatra.speech.validate or synthesis refuses the text as too long, follow the refusal message: split at natural section or scene boundaries, list every part upfront, and submit the parts in order. If a dialogue request is refused as too long, split it at scene or turn boundaries as the refusal message says.

Create, deliver, and record

Lead the reply with the result and format the user asked for; the production card, route choice, and cost notes follow it and never replace it.

On first use or after the package version changes, make a best-effort non-billable installation registration (beatra.installations.register). Its failure never blocks the episode workflow.

Invoke every remote Beatra tool only through the bundled scripts/mcp_client.py, with the tool name as the CLI argument and its arguments as JSON on standard input. Do not configure or call a host Beatra Connector, and do not use REST/OpenAPI as a fallback. Submit each approved logical paid request once under its stable request ID, record its returned task ID immediately, and poll that task with beatra.tasks.get until it is terminal.

For optional finishing (beatra.music.generate, beatra.audio.generate_sfx with prompt, duration_seconds 0.5–30.0, optional loop=true, and beatra.audio.process normalize_loudness -14 LUFS, trim, fade — up to 60 min / 300 MB, min result >= 1 s, 50 calls/day, one operation per call), submit each approved call once under its own frozen card and ID. See music editing and audio.

Deliver returned episode audio with its real artifact ID, duration, MIME type, sample rate when present, speech_start_seconds / speech_end_seconds / integrated_loudness_lufs when present, free output.subtitles (including dialogue segments[] when using beatra.speech.synthesize_dialogue), resolved model, usage, and returned billing facts. Re-time chapter timestamps from real returned durations and subtitle segment timing (zero extra subtitle cost). Also deliver three title options, the episode description, and show notes; the closing only recaps what was said. Deliver any music bed, SFX stinger/ambient loop, or finished audio clip honestly as its own artifact — never claim unmixed stems are a mixed multi-track final episode. When the host can listen, review pacing, pronunciations, speaker turns, and loudness fit; otherwise make the outstanding listening review explicit. Update the user's show ledger if permitted, or hand the factual record back. For a lost task ID or error recovery, follow voice, delivery, and recovery and speech, voices, cloning, and audio finishing.

References by task

Updates and security

This Skill runs only the files installed with it and leaves them unchanged while it runs; it never downloads or executes other code. It talks to Beatra's official services over HTTPS, and a file you choose to upload goes only to the one-time upload address Beatra issues for it.

Once a day at most, the bundled client reads Beatra's public version list. When the store or tool you installed from has a newer version, it prints one line with the command that updates it there. Updating stays your choice, through that store or tool. See updates and security.

slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack."Stars: 198kUpdated Claude CodeCodexVerifiedInstall
ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read.Stars: 196kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report.Stars: 196kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt.Stars: 196kUpdated Claude CodeCodexVerifiedInstall