> ## Documentation Index
> Fetch the complete documentation index at: https://composio-preview.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Inworld AI

> Inworld AI provides native APIs for speech synthesis, speech recognition, voice management, model discovery, embeddings, knowledge, and router management.

![Inworld AI logo](https://mintcdn.com/composio-preview/tIuAovF9-7pXq2wA/images/toolkits/inworld_ai.svg?fit=max\&auto=format\&n=tIuAovF9-7pXq2wA\&q=85\&s=fb7ccf1d9a3f246a89d5e1e6ab0fb38c)

* **Category:** artificial intelligence
* **Auth:** API\_KEY
* **Composio-managed OAuth available?** N/A
* **Tools:** 6
* **Triggers:** 0
* **Slug:** `INWORLD_AI`
* **Version:** 20261101\_00

## Tools

> [accordion] Create Embeddings
>
> **Slug:** `INWORLD_AI_CREATE_EMBEDDINGS`
>
> Generate one complete embedding vector per input text with an Inworld-hosted embedding model. Batches can produce large responses.
>
> #### Input Parameters
>
> **texts** `array` *required*
>
> Texts to embed, in output order. A call accepts 1-16 texts; each returned vector can be large.
>
> **model\_id** `string`
>
> Embedding model ID. The default model was verified against the live API.
>
> **service\_provider** `string`
>
> Embedding service provider identifier. Use SERVICE\_PROVIDER\_INWORLD with the verified default model.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Get Voice Preview
>
> **Slug:** `INWORLD_AI_GET_VOICE_PREVIEW`
>
> Generate the provider-selected, unmetered MP3 preview for a voice and return it as a downloadable file.
>
> #### Input Parameters
>
> **model\_id** `string` *required*
>
> TTS model used to select the preview.
>
> **voice\_id** `string` *required*
>
> Voice ID from LIST\_VOICES. System voice IDs are valid for previews.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] List Models
>
> **Slug:** `INWORLD_AI_LIST_MODELS`
>
> List Inworld-supported models and the pricing, modality, context, and output limits needed to choose a model. The endpoint returns the complete model catalog and can produce a moderately large response.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] List Voices
>
> **Slug:** `INWORLD_AI_LIST_VOICES`
>
> Return one page of voices available for preview or speech synthesis, with optional filtering, language selection, and sorting. The filter and languages parameters are mutually exclusive.
>
> #### Input Parameters
>
> **cursor** `string`
>
> Continuation token returned by a previous call; omit for the first page.
>
> **filter** `string`
>
> Optional AIP-160 filter, for example source = "IVC" for workspace-owned voices or source = "SYSTEM" for system voices. Cannot be combined with languages.
>
> **order\_by** `string`
>
> Comma-separated voice fields to sort by, with optional desc after a field.
>
> **languages** `array`
>
> Optional BCP-47 language codes to include, such as en-US. Cannot be combined with filter.
>
> **page\_size** `integer`
>
> Maximum voices to return in this page (1-2000). Smaller pages reduce response size.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Synthesize Speech
>
> **Slug:** `INWORLD_AI_SYNTHESIZE_SPEECH`
>
> Synthesize short text with an Inworld voice and return the complete audio as a downloadable file. Long text can exceed tool payload limits; prefer short text and compressed MP3 output.
>
> #### Input Parameters
>
> **text** `string` *required*
>
> Text to speak (1-2000 characters). Prefer short text because generated base64 audio can exceed Composio/runtime payload limits.
>
> **bit\_rate** `integer`
>
> Optional encoded-audio bit rate supported by the selected encoding and model.
>
> **language** `string`
>
> Optional BCP-47 language code when language should be selected explicitly.
>
> **model\_id** `string` *required*
>
> TTS model ID, for example inworld-tts-2.
>
> **voice\_id** `string` *required*
>
> Voice ID from LIST\_VOICES. System voice IDs are accepted by synthesis.
>
> **speaking\_rate** `number`
>
> Optional speech-speed multiplier supported by the selected model.
>
> **audio\_encoding** `string ("MP3" | "LINEAR16" | "OGG_OPUS" | "MULAW" | "ALAW" | "FLAC")`
>
> Output audio encoding. MP3 is recommended to minimize transport size.
>
> **sample\_rate\_hertz** `integer`
>
> Optional output sample rate in hertz supported by the selected encoding and model.
>
> **enhance\_generation** `boolean`
>
> Whether to enable the provider's generation enhancement when supported by the selected model.
>
> **apply\_text\_normalization** `string ("ON" | "OFF" | "APPLY_TEXT_NORMALIZATION_UNSPECIFIED")`
>
> Text-normalization mode. ON expands values such as numbers and abbreviations, OFF reads text as written, and APPLY\_TEXT\_NORMALIZATION\_UNSPECIFIED lets Inworld decide.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Transcribe Audio
>
> **Slug:** `INWORLD_AI_TRANSCRIBE_AUDIO`
>
> Synchronously transcribe a short audio file and return the transcript, with word timestamps and usage when available. Prefer compressed input to limit base64 payload growth.
>
> #### Input Parameters
>
> **prompts** `array`
>
> Optional words or phrases that help the model recognize expected names or terminology.
>
> **language** `string`
>
> Optional language code such as en; omit to use provider detection or default behavior.
>
> **model\_id** `string`
>
> Synchronous STT model ID. The default was live-verified; groq/whisper-large-v3 is also supported for synchronous use.
>
> **audio\_file** `object` *required*
>
> Short audio file to transcribe. Prefer compressed MP3 and avoid large files because the API embeds audio as base64 JSON.
>
> **audio\_encoding** `string ("AUTO_DETECT" | "LINEAR16" | "MP3" | "OGG_OPUS" | "FLAC")`
>
> Encoding of audio\_file; it must match the uploaded audio bytes. Use AUTO\_DETECT for self-describing formats when the encoding is unknown.
>
> **sample\_rate\_hertz** `integer`
>
> Sample rate in hertz for raw audio; omit for self-describing container formats such as MP3.
>
> **number\_of\_channels** `integer`
>
> Channel count for raw audio; omit for self-describing container formats such as MP3.
>
> **include\_word\_timestamps** `boolean`
>
> Whether to include per-word timestamps in the transcription.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not
