> ## Documentation Index > Fetch the complete documentation index at: https://docs.anyreach.ai/llms.txt > Use this file to discover all available pages before exploring further. # Resolve Greeting Audio Playback Url > Mint a short-lived presigned GET URL to re-preview greeting audio. Recomputes the content fingerprint from ``request.text`` + the version's TTS config (the same way the runtime does), then presigns the stored object. Returns 404 if no recording exists for that exact text+voice (e.g. the text was edited but not regenerated). The bucket and org come from trusted server state, so there is no caller-supplied location. For an agent with a randomized voice pool, *every* pool voice must have a recording before this reports success: a partial set still plays, but the calls that draw a missing voice fall back to live TTS, and the builder saying "pre-generated" while some calls are not would be the lie this endpoint exists to prevent. One presigned URL is returned per pool voice so each can be auditioned. ## OpenAPI ````yaml /openapi.json post /core/agents/{id}/versions/{version}/starting-message/audio/playback-url openapi: 3.1.0 info: title: Anyreach API version: 1.0.0 description: >- REST API for the Anyreach platform. Every endpoint is scoped to one organization. Authenticate with a bearer token and, if your token can reach more than one organization, name the one you mean with the `X-Anyreach-Org` header. See [Authentication](https://docs.anyreach.ai/api-reference/authentication). This reference is generated from the running service, so the request and response shapes here are the ones the API actually enforces. servers: - url: https://api.anyreach.ai description: Production security: - bearerAuth: [] tags: - name: Agents description: >- Create and configure the agent that handles a conversation, and manage its versions. An agent's behaviour lives in a version; publishing a version is what puts it on the air. - name: Conversations description: >- Create, list, retrieve, and message conversations between your users and Anyreach agents. - name: Conversation attachments description: >- Upload files into a conversation and read them back. Uploads are two-step: presign, PUT to the returned URL, then mark the attachment complete. - name: Conversation events description: >- The ordered event stream behind a conversation — turns, tool calls, transfers, and status changes. - name: Knowledge bases description: >- A knowledge base (a `dataset` in the API) is a collection of embedded content an agent can retrieve from. - name: Knowledge base sources description: Attach and detach sources to a knowledge base. - name: Sources description: >- A source is a file or URL whose content is ingested, chunked, and embedded. - name: Phone numbers description: >- Buy, list, and release phone numbers, and point each one at the agent that should answer it. - name: Trunks description: >- SIP trunks for bringing your own carrier. A trunk carries calls between your carrier and Anyreach. - name: Inbound routing description: >- Rules that decide which agent answers an inbound call, evaluated before the call connects. - name: Agent assist description: >- Look up the live-assist context for a number so a human agent can see what the AI agent knows. - name: Human handoff description: >- Connections to the help desk a conversation escalates into, and the test endpoint that proves one works. - name: Email domains description: >- Register a sending domain and verify its DNS before an agent can send from it. - name: Email addresses description: >- The addresses on a verified domain, and the agent that handles mail arriving at each one. - name: Unsubscribes description: >- The suppression list. An address on this list is never emailed again until it is removed. - name: Web widgets description: >- The embeddable chat and voice widget: its appearance, its allowed domains, and the agent behind it. - name: Campaigns description: >- An outbound campaign dials a contact list with a given agent on a given schedule. Start, pause, and close control its lifecycle. - name: Campaign contacts description: >- The people a campaign dials. Contacts can be added while the campaign is running. - name: Campaign configurations description: >- Reusable dialing settings — calling windows, retry policy, timezone handling — that campaigns reference. - name: Campaign stats description: >- Per-campaign outcome counters, and the organization-wide cap on how many calls run at once. - name: Custom data description: >- Your own tables, queryable by an agent mid-conversation. Define a table, load rows, and expose read-only stored queries as agent tools. - name: Metrics description: >- The definitions an LLM judge scores a conversation against. Lint and preview a metric before you run it at scale. - name: Evaluations description: A scorecard binding a set of metrics to a population of conversations. - name: Evaluation runs description: >- One execution of an evaluation over a conversation set, and the results it produced. - name: Workflows description: >- Automations built from steps. A workflow is edited as a version and goes live when a version is published and set live. - name: Workflow triggers description: >- What starts a workflow — a schedule, a webhook, or a point in an agent conversation. Triggers are versioned alongside the workflow. - name: Workflow executions description: >- Run a workflow and read back what happened. Executions are synchronous by default; use the async endpoint for long-running work. - name: Workflow credentials description: >- Secrets a workflow step authenticates with. Values are write-only — they can be set and rotated, never read back. paths: /core/agents/{id}/versions/{version}/starting-message/audio/playback-url: post: tags: - Agents summary: Resolve Greeting Audio Playback Url description: |- Mint a short-lived presigned GET URL to re-preview greeting audio. Recomputes the content fingerprint from ``request.text`` + the version's TTS config (the same way the runtime does), then presigns the stored object. Returns 404 if no recording exists for that exact text+voice (e.g. the text was edited but not regenerated). The bucket and org come from trusted server state, so there is no caller-supplied location. For an agent with a randomized voice pool, *every* pool voice must have a recording before this reports success: a partial set still plays, but the calls that draw a missing voice fall back to live TTS, and the builder saying "pre-generated" while some calls are not would be the lie this endpoint exists to prevent. One presigned URL is returned per pool voice so each can be auditioned. operationId: resolveGreetingAudioPlaybackUrl parameters: - name: id in: path required: true schema: type: string title: Id - name: version in: path required: true schema: type: string title: Version - name: x-anyreach-org in: header required: false schema: anyOf: - type: string - type: 'null' description: Organization ID for user PATs (pat_ prefix tokens) title: X-Anyreach-Org description: Organization ID for user PATs (pat_ prefix tokens) requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/GreetingAudioRequest' responses: '200': description: Successful Response content: application/json: schema: $ref: '#/components/schemas/GreetingAudioResponse' example: recordings: - voice_id: 3fa85f64-5717-4562-b3fc-2c963f66afa6 playback_url: https://example.com/resource playback_url: https://example.com/resource '422': description: Validation Error content: application/json: schema: $ref: '#/components/schemas/HTTPValidationError' security: - HTTPBearer: [] components: schemas: GreetingAudioRequest: properties: text: type: string title: Text force: type: boolean title: Force default: false type: object required: - text title: GreetingAudioRequest description: >- Request to pre-generate (or re-preview) static-greeting audio. ``text`` is supplied by the caller so it works on unsaved drafts (mirrors ``suggest-action-grid``). The TTS voice is read from the saved version's configured TTS model, and the stored object is content-addressed over ``text`` + that config, so the same body identifies the same recording. GreetingAudioResponse: properties: recordings: items: $ref: '#/components/schemas/GreetingAudioRecording' type: array title: Recordings playback_url: type: string title: Playback Url description: |- The primary voice's recording (pool entry 0 when there is a pool). Serialized alongside ``recordings`` so a client that predates the per-voice list keeps working — including across a deploy where the backend rolls out ahead of the builder. readOnly: true type: object required: - recordings - playback_url title: GreetingAudioResponse description: |- Short-lived presigned URLs for playing/previewing greeting audio. An agent with a randomized voice pool has one recording per pool voice — any of them can open a real call, so all of them are auditionable. HTTPValidationError: properties: detail: items: $ref: '#/components/schemas/ValidationError' type: array title: Detail type: object title: HTTPValidationError GreetingAudioRecording: properties: voice_id: anyOf: - type: string - type: 'null' title: Voice Id playback_url: type: string title: Playback Url type: object required: - playback_url title: GreetingAudioRecording description: One pre-generated greeting recording, in one voice. ValidationError: properties: loc: items: anyOf: - type: string - type: integer type: array title: Location msg: type: string title: Message type: type: string title: Error Type type: object required: - loc - msg - type title: ValidationError securitySchemes: bearerAuth: type: http scheme: bearer description: Personal access token or organization API key. See Authentication. ````