switch

Generate Audio

generate_audio

Generate spoken audio from text: narration, a voiceover, a read-aloud script, or a multi-voice dialogue. Pass text (up to 2048 chars) — the words to be spoken. To speak in one of YOUR saved voices, pass voice with the voice NAME (or id): users speak plain language and never know ids, so resolve the name yourself (the voice tool, action "list", shows every saved voice) and never ask the user for an id. Reference voices, trained clones and preset voices are all routed correctly by kind. To match a voice instantly from a clip instead, pass reference_audio_url (a short clip) or up to 3 reference_audio_urls and address them as @Audio1, @Audio2, @Audio3 in the text for dialogue. Alternatively pass image_url to voice a scene from a picture (cannot combine with reference audio). Optional speech_rate (-50..100), pitch (-12..12), loudness (-50..100). Returns a playable audio_url, duration_seconds, and generation_id (also saved to your library).

Input Schema

TableJSON Schema

Name	Required	Description
`text`	Yes	The words to speak / narrate / perform. Max 2048 chars. For dialogue, address voices as @Audio1, @Audio2, @Audio3.
`pitch`	No	Optional. Pitch, -12 to 12. 0 is normal.
`voice`	No	Optional. A saved voice — pass its NAME (or id); it is resolved and routed by kind automatically. Omit for a natural default voice.
`format`	No	Optional output format. Default mp3.
`loudness`	No	Optional. Loudness, -50 (quieter) to 100 (louder). 0 is normal.
`image_url`	No	Optional. Voice a scene from a picture. Cannot be combined with reference audio.
`speech_rate`	No	Optional. Speaking speed, -50 (slower) to 100 (faster). 0 is normal.
`reference_audio_url`	No	Optional. A short clip URL to instantly match that voice.
`reference_audio_urls`	No	Optional. Up to 3 reference clip URLs for multi-voice dialogue.

TDQS

A4.7/5.0

Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

The description discloses far beyond the readOnlyHint=false annotation: it explains side effects (generation is saved to library), parameter constraints (text up to 2048 chars, speech_rate, pitch, loudness ranges), voice routing behavior (reference voices, trained clones, presets routed by kind), and dialogue addressing with @Audio1/@Audio2/@Audio3. This gives the agent a clear model of what happens on invocation.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is a single dense paragraph but every sentence carries useful information for a 9-parameter tool. It front-loads the core purpose, then logically covers voice options, alternates, and return fields. It could be slightly improved with bullet points or section breaks, but length is justified by complexity.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a tool with no output schema, the description explicitly states the return values (playable audio_url, duration_seconds, generation_id) and the side effect of saving to the library. It also covers the full range of input modes (text, saved voice, reference audio, image) and their constraints, making the tool self-contained for the agent.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters5/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Although schema coverage is 100%, the description adds significant meaning: voice can be a name or id and is resolved automatically, reference_audio_urls enable multi-voice dialogue via @Audio tags, image_url cannot combine with reference audio, and numeric parameter ranges are clarified. This goes well beyond the schema property descriptions.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description opens with a specific verb and resource: 'Generate spoken audio from text,' then lists concrete use cases (narration, voiceover, read-aloud script, multi-voice dialogue). This clearly distinguishes it from sibling generation tools like generate_image or generate_video, which produce different media types.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

It provides explicit usage modes and alternatives: using a saved voice by name, matching from a reference clip, or voicing an image, with the note that image_url cannot combine with reference audio. It also gives operational guidance (resolve voice names yourself, never ask the user for an id) and cross-references the voice tool's list action. It lacks an explicit 'use this instead of X' statement for sibling tools, but internal alternatives are well covered.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All switch tools Inspect

Try in Browser

RayrunBuild, deploy, and operate MCP servers without becoming an infrastructure team.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

TDQS

A3.5/5.0

Disambiguation2/5

Several tools occupy nearly identical semantic ground: apply_iphone_realism and apply_ugc both describe casual phone-shot looks, upload_media and upload_reference_asset both accept uploads, and analyze_video overlaps heavily with analyze_video_report. The many apply_* style tools are essentially one tool parameterized by style, so agents can easily select the wrong one.

Naming Consistency4/5

Most tools follow a clear verb_noun snake_case pattern such as generate_image, list_my_videos, get_editor_run, and upscale_video. A few outliers like voice, talking_avatar_video, and video_to_prompt do not use the same verb-first convention, but they are still readable and do not create significant confusion.

Tool Count1/5

At 55 tools, the surface is far beyond what is appropriate for an MCP server; many of these be collapsed or parameterized, especially the 10 apply_* style wrappers and several overlapping upload/status helpers. Even for a broad media platform, this scale forces a huge context window and makes selecting the right tool impractical.

Completeness2/5

The surface covers generation, media display, video analysis, and Editor workflows well, but there are obvious gaps in library lifecycle management: move_asset and create_folder are referenced in tool descriptions without being exposed, and there is no clean way to delete or reorganize media assets. Agents following the descriptions will try to call tools that do not exist.

Resources

Need Help?