# Voice

> Voice catalog and library, and the voice settings of a flow.

Voice catalog and library, and the voice settings of a flow.

## `flow_voice_get`

**Get the voice of a flow** · Read only · Listed to the agent from the start

Voice configuration of a flow's draft (`settings.voiceConfig`: stt, tts, timings, messages) plus the readable voice name (`voiceName`).

| Argument | Type | Required | Description |
| --- | --- | --- | --- |
| `client_id` | string | Yes |  |
| `flow_id` | string | Yes |  |

## `flow_voice_set`

**Set the voice of a flow** · Changes data · can overwrite or remove · Reached with `dakso_explore`

Applies voice settings to the draft of a voice flow (merged into settings.voiceConfig, always with enabled: true). It does not publish: the change enters calls when the flow is published. Only what changes is sent (empty / 0 = leave). `tts_provider`: dakso-realtime, dakso-essential or elevenlabs; `stt_provider`: dakso-standard, dakso-conversational or dakso-multilingual (see `voice_models_list`). ElevenLabs needs `voice` and a connected key. Changing provider drops the previous voice and model. `speed` 0.7-1.2; `max_call_duration` and `idle_timeout` in seconds. Returns {applied, draft: true, note}.

| Argument | Type | Required | Description |
| --- | --- | --- | --- |
| `client_id` | string | Yes |  |
| `flow_id` | string | Yes |  |
| `descripcion` | string | No | One short sentence: what this call changes and why (10-200 chars, no personal data). |
| `idle_timeout` | integer | No | Default: `0`. |
| `language` | string | No | Default: `""`. |
| `max_call_duration` | integer | No | Default: `0`. |
| `speed` | number | No | Default: `0`. |
| `stt_model` | string | No | Default: `""`. |
| `stt_provider` | string | No | Default: `""`. |
| `tts_model` | string | No | Default: `""`. |
| `tts_provider` | string | No | Default: `""`. |
| `voice` | string | No | Default: `""`. |
| `voice_name` | string | No | Default: `""`. |

## `voice_catalog_search`

**Search voices** · Read only · Reached with `dakso_explore`

Searches the voice catalog: Dakso voices (dakso-realtime, dakso-essential) and the organization's ElevenLabs account (elevenlabs). Company voices are in `voice_library_list`. `provider` empty = all. `gender` = female|male|neutral; `accent` (e.g. "latin", "mexican") and `query` (name or description) filter. `source`: "account", "library" (public ElevenLabs library; applying it with `flow_voice_set` adds it to the account), "cloned" or empty. `limit` up to 12. It does not clone voices: that is done in the builder (/settings/voices?tab=clone).

| Argument | Type | Required | Description |
| --- | --- | --- | --- |
| `accent` | string | No | Default: `""`. |
| `gender` | string | No | Default: `""`. |
| `language` | string | No | Default: `"es"`. |
| `limit` | integer | No | Default: `8`. |
| `provider` | string | No | Default: `""`. |
| `query` | string | No | Default: `""`. |
| `source` | string | No | Default: `""`. |

## `voice_library_add`

**Add a voice to the library** · Changes data · can overwrite or remove · Reached with `dakso_explore`

Adds a voice to the company library (it touches no flow). If it is a public ElevenLabs library voice not yet in the account, it is first added to the account (requires a connected ElevenLabs key). Requires the implementer role. It does not clone voices: that is done in the builder (/settings/voices?tab=clone).

| Argument | Type | Required | Description |
| --- | --- | --- | --- |
| `provider` | string | Yes |  |
| `voice_id` | string | Yes |  |
| `accent` | string | No | Default: `""`. |
| `descripcion` | string | No | One short sentence: what this call changes and why (10-200 chars, no personal data). |
| `gender` | string | No | Default: `""`. |
| `language` | string | No | Default: `"es"`. |
| `name` | string | No | Default: `""`. |

## `voice_library_list`

**List the voice library** · Read only · Reached with `dakso_explore`

The company's voice library (the voices the team chose) and the default voice. Look here first when recommending a voice. Returns {voices: [{provider, voiceId, name, language, accent, gender, source, defaultModel, isDefault, ...}], default: {provider, voiceId, defaultModel, sttProvider, sttModel} | null}.

No arguments.

## `voice_library_remove`

**Remove a voice from the library** · Changes data · can overwrite or remove · Reached with `dakso_explore`

Removes a voice from the company library (it is not deleted from the provider or from flows). If the voice is the default or used by published flows, it is not removed: it returns {removed: false, confirmRequired: true, usedBy, wasDefault}. Repeating with `confirm=True` removes it anyway, but that variant is not run by the agent: it returns a link to the admin screen where a person confirms it. Requires the implementer role.

| Argument | Type | Required | Description |
| --- | --- | --- | --- |
| `provider` | string | Yes |  |
| `voice_id` | string | Yes |  |
| `confirm` | boolean | No | Default: `false`. |
| `descripcion` | string | No | One short sentence: what this call changes and why (10-200 chars, no personal data). |

## `voice_library_set_default`

**Set the default voice** · Changes data · can overwrite or remove · Reached with `dakso_explore`

Sets the company's default voice (with its TTS model and, optionally, the default STT). New voice flows start with this voice; existing ones do not change. If the voice is not in the library it is added (`voice_name` gives its name). Requires the implementer role.

| Argument | Type | Required | Description |
| --- | --- | --- | --- |
| `provider` | string | Yes |  |
| `voice_id` | string | Yes |  |
| `descripcion` | string | No | One short sentence: what this call changes and why (10-200 chars, no personal data). |
| `model` | string | No | Default: `""`. |
| `stt_model` | string | No | Default: `""`. |
| `stt_provider` | string | No | Default: `""`. |
| `voice_name` | string | No | Default: `""`. |

## `voice_models_list`

**List voice models** · Read only · Reached with `dakso_explore`

Supported voice models: speech-to-text (dakso-standard, dakso-conversational, dakso-multilingual) and text-to-speech (dakso-realtime, dakso-essential, and elevenlabs = the organization's ElevenLabs account), with latency, quality and cost (1-3, 3 = fastest, best, priciest), languages, a display `label` and each default. `kind` = "stt" or "tts" filters; `provider` filters.

| Argument | Type | Required | Description |
| --- | --- | --- | --- |
| `kind` | string | No | Default: `""`. |
| `provider` | string | No | Default: `""`. |
