Voiceover & Narration Studio
Choose or reuse a voice, turn scripts into ready-to-edit narration, plan ordered long-form or multilingual audio, and create a reusable custom voice.
Voiceover & Narration Studio
Choose the smallest voice workflow that completes the user's approved text: one voiceover, an ordered long-form production, supplied multilingual versions, an authorized reusable voice, or broadcast finishing on voice tracks and clone samples. Keep this Studio as the coordinating route when a request mixes those outcomes or continues across sessions.
Scope and routes
At the entrance, lock the medium first: long narration, multilingual, or brand
clone. Route an audiobook to ai-audiobook-narration, a podcast episode to
ai-podcast-voiceover, and a short social or promo read to
short-form-voiceover-audio.
Reuse every decision already present in the conversation. Route a focused, standalone request to the narrower or specialized package when it is available:
| User outcome | Best route |
|---|---|
| One short-video, ad, social, or promo read | short-form-voiceover-audio |
| An audiobook or other ordered long-form narration | ai-audiobook-narration |
| A single-host podcast episode or recurring show | ai-podcast-voiceover |
| Multilingual dubbing with target-language scripts already supplied | ai-multilingual-dubbing |
| Dedicated cloning or voice deletion | voice-cloning-studio |
| Multi-speaker dialogue, voice conversion, transcription, sound effects, or stem separation | beatra-ai-voice-studio |
| Long narration, multilingual reads, brand cloning, sample isolation, or loudness normalization | Continue here and maintain one production ledger |
This Studio creates speech from approved text, clones authorized voices,
pre-cleans noisy clone samples (isolate_voice), and runs loudness
normalization (normalize_loudness). It does not make finished video,
lip-sync speakers, or run operations outside scripts/mcp_client.py tools
(such as multi-speaker dialogue, speech-to-speech conversion, transcription,
sound effects, stem separation, or voice deletion).
Read intent and routing and the shared speech and audio guide when planning routes or verifying capabilities.
Inputs and defaults
Hard inputs include consent status. Never imitate a voice without authorization.
For synthesis, require:
- final supplied text, or permission to prepare a speakable version (manually or via Smart Voice Director) and show it before paid synthesis;
- an exact current
voice_id, looked up fresh withbeatra.voices.listeven when named earlier in the thread; - a target language only when the user supplies one; otherwise omit
languageand never interrupt just to ask for it.
For voice cloning, also require a clean single-speaker sample, a display name, and the user's plain confirmation that it is their own voice or that the speaker explicitly authorized cloning—confirmed before any local upload.
Use restrained defaults unless destination delivery requires otherwise:
model: "auto", format: "mp3", speed: 1.0, volume: 1.0, pitch: 0,
no emotion, and no sample_rate. Preserve the user's simplified/traditional
characters and dialect wording verbatim unless asked to change them. Never
hand-strip / pause markers or invent expressive bracket tags; let Smart Voice
Director compile natural pacing and delivery cues when dramatic coaching helps.
Golden path
- Identify one route or an ordered ledger of routes (clone, script direction,
chapters, languages, and loudness finishing stay distinct). Check
scripts/mcp_client.py toolsso every planned step exists on this account. - Run a best-effort non-billable
beatra.installations.registeron first use or after a version change (registration failure never blocks voice work). - Cast by live lookup: filter
beatra.voices.listusing its 11 capability and style dimensions (category,language,gender,age_group,use_case,emotion,supports_context,supports_direction,supports_performance_tags,supports_dialogue,supports_speech_to_speech), confirm exactvoice_idandstatus == "ready", and audition free with returnedpreview_urlorbeatra.voices.preview({"voice_id": "..."}). Keepbeatra.voices.updateready to rename or pin a reusable series/brand default. - Check live model cards and estimates via
beatra.models.listfor the exact capability needed (speech_script,text_to_speech,voice_clone,audio_voice_isolation, oraudio_process). Never hardcode credit numbers, upstream vendor names, or internal model IDs. - Prepare text accurately before paid synthesis:
- Use paid Smart Voice Director (
beatra.speech.prepare_script, after showing the six-field production card and obtaining approval) when pacing, pauses, or performance cues would improve expressive reads. Inspectdirectedandscripts[]: passscripts[i].synthesisunchanged with a freshclient_request_idinto synthesis, handlesynthesis: nullfromwarnings, and note that re-callingprepare_scriptwithin 48 hours with identical text/instructions/language/context while changing onlyvoiceormodelrecompiles at zero extra charge (subject to account balance covering initial estimate and max 2 concurrentprepare_scripttasks). Never callprepare_scriptgenerally free. - Always run free
beatra.speech.validate(estimated_duration_seconds, character bounds, control fit) beforebeatra.speech.synthesize. Request free alignedSRT/VTTsubtitles viasubtitle_enable: truewhen captions or edit markers help (never passsubtitlesonspeech.synthesize).
- Use paid Smart Voice Director (
- Show the six-field production card before the first paid step (
prepare_script,isolate_voice,voices.clone,speech.synthesize, ornormalize_loudness): Route & Model, Exact Scope, Price & Basis, Duration / Quantity, Pay Mode & Call Count, and If stop (stating clearly what free deliverables the user keeps at zero extra charge, such as prepared drafts, voice shortlist/previews, or validation checks). Wait for explicit confirmation on clones, Smart Voice Director scripts, noisy-sample isolation, loudness normalization, pilots, or remainders; only an explicit single-read synthesis request that already names every frozen parameter avoids redundant asking. Use the bundledscripts/mcp_client.pyfor every Beatra operation:
python3 scripts/mcp_client.py call <tool-name>
Pass one JSON object to standard input, and use
python3 scripts/mcp_client.py upload <path> --mime-type <exact MIME> for a
local sample. Then wait for returned results, run serialized finishing calls one
at a time (normalize_loudness, e.g. -14 LUFS, <= 60 min / <= 300 MB / >= 1 s
output, 50 calls/day limit), and deliver only returned artifacts and billing
facts in order. Generate one opaque local client_request_id per paid call,
store the ledger locally before sending, and poll only the returned task_id
via beatra.tasks.get.
Never configure or call a host Beatra Connector. Never use REST/OpenAPI as a
fallback, never imitate the upload protocol, and never claim a paid step
(prepare_script, isolate_voice, normalize_loudness) is free.
Voice cloning and consent
Mandatory consent gate before any local sample upload: obtain plain confirmation
that the audio is the user's own voice or explicitly authorized by the speaker.
If a sample carries background noise, offer paid pre-cleaning via
beatra.audio.process (operation: "isolate_voice", source field, 4.6 s to
5 min, <= 100 MB, priced against {"capability": "audio_voice_isolation"}).
Then select the clone model as described in
speech and audio guide §5; never mention
vendor names or internal IDs. Show the six-field production card with the live
clone price before voices.clone; after the user approves it, submit once and
handle any refusal per
billing, errors, and recovery.
When a clone result or a listed voice has status pending_activation, follow §5
Step E of speech and audio guide before using
the voice. Keep any spoken test reading as a separate synthesis card after
cloning succeeds.
Recovery and balance rules
Never treat beatra.wallet.get as a required gate before paid submission, and
never ask the user to confirm they have enough credits. Only call
beatra.wallet.get when asked about current balance/fit, or
beatra.wallet.ledger when asked about deductions; both are read-only. If a
model card returns top_up, list its tiers verbatim in returned order without
ranking or picking one.
On insufficient_balance, relay Beatra's official message and exact topup_url
verbatim, stop until the user confirms topping up, and then replay the exact
frozen payload with the same client_request_id. If the refusal names another
way forward, present it too and follow
billing, errors, and recovery.
Recover uncertain/lost tasks through paginated beatra.tasks.list
(next_cursor) plus beatra.tasks.get; any changed text, voice, language,
model, or finishing parameter requires a new card, approval, and
client_request_id.
References by task
- Read speech and audio guide for cross-tool limits, Smart Voice Director payloads, clone model choice, and card rules.
- Read intent and routing for the Studio matrix, minimum inputs, free planning, and paid pilot boundaries.
- Read voice casting and delivery
for 11-dimension voice lookup, live model/language checks, weighted-character
estimates,
validate+synthesizeJSON, andnormalize_loudness. - Read long-form and multilingual production for route ledgers, pilot coverage, ordered batches, and interruption recovery.
- Read voice cloning and review for
consent,
isolate_voicepre-clean, sample upload,voices.clone/voices.update, and separate spoken proof. - Read bundled MCP Client diagnostics only for bundled-client troubleshooting.
- Read installation and authentication, installation registration, tasks and results, billing, errors, and recovery, and uninstall and disconnect when active.
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.
Related skills
- Downloads
- 1.1kAll time
- Security audit
- Review
- Last updated
- 2d ago
- Current version
- v0.3.1
- License
- MIT-0
