Plivo Audio Stream API

asyncapi: 3.0.0
info:
  title: Plivo Audio Streaming WebSocket API
  version: '1.0.0'
  description: >-
    The Plivo Audio Streaming API delivers near real-time raw audio from active
    Plivo voice calls to a customer-operated WebSocket server, and (when
    bidirectional streaming is enabled) accepts audio and control events from
    the server back into the live call. The customer-operated WebSocket
    endpoint is declared in the Plivo XML response that controls the call,
    using the `<Stream>` element. Plivo opens a WSS connection to that endpoint
    and exchanges JSON text frames following the Plivo Audio Streaming event
    protocol.


    Audio formats supported on the wire: `audio/x-l16;rate=8000`,
    `audio/x-l16;rate=16000`, and `audio/x-mulaw;rate=8000`. Audio payloads in
    each direction are base64-encoded.


    Stream lifecycle notifications (stream stopped, stream timeout, stream
    failed) are delivered by Plivo over a separate HTTP status callback to the
    `statusCallbackUrl` configured on the `<Stream>` element and are not part
    of the WebSocket event protocol.
  contact:
    name: Plivo
    url: https://www.plivo.com/docs/
  license:
    name: Proprietary
    url: https://www.plivo.com/legal/
  externalDocs:
    description: Plivo Audio Streaming - Stream Event Protocol
    url: https://www.plivo.com/docs/voice-agents/audio-streaming/concepts/stream-event-protocol

defaultContentType: application/json

servers:
  customer-websocket:
    host: 'yourserver.example.com'
    pathname: /audiostream
    protocol: wss
    description: >-
      Customer-operated WebSocket Secure (WSS) endpoint declared inside the
      Plivo `<Stream>` XML element. Plivo connects from its voice
      infrastructure to this URL when the call reaches the `<Stream>`
      instruction. The host and path shown here are illustrative — the actual
      value is whatever the customer publishes in the XML.
    externalDocs:
      description: The Stream XML element
      url: https://www.plivo.com/docs/voice/xml/audiostream

channels:
  audioStream:
    address: /
    description: >-
      Single bidirectional JSON-over-WebSocket channel established when Plivo
      connects to the customer's WSS endpoint declared in `<Stream>`. All
      events listed below flow over this same connection.
    messages:
      start:
        $ref: '#/components/messages/StartEvent'
      media:
        $ref: '#/components/messages/MediaEvent'
      dtmf:
        $ref: '#/components/messages/DtmfEvent'
      playedStream:
        $ref: '#/components/messages/PlayedStreamEvent'
      clearedAudio:
        $ref: '#/components/messages/ClearedAudioEvent'
      playAudio:
        $ref: '#/components/messages/PlayAudioEvent'
      checkpoint:
        $ref: '#/components/messages/CheckpointEvent'
      clearAudio:
        $ref: '#/components/messages/ClearAudioEvent'
      sendDTMF:
        $ref: '#/components/messages/SendDTMFEvent'

operations:
  receiveStart:
    action: receive
    channel:
      $ref: '#/channels/audioStream'
    summary: Stream started
    description: >-
      Sent by Plivo to the customer WebSocket server when the WSS connection
      is established and audio streaming for the call begins. Carries call,
      stream, account, track, and media-format metadata.
    messages:
      - $ref: '#/channels/audioStream/messages/start'
  receiveMedia:
    action: receive
    channel:
      $ref: '#/channels/audioStream'
    summary: Inbound audio chunk
    description: >-
      Sent by Plivo to deliver a chunk of base64-encoded raw audio (~20ms per
      chunk) from the configured tracks of the live call.
    messages:
      - $ref: '#/channels/audioStream/messages/media'
  receiveDtmf:
    action: receive
    channel:
      $ref: '#/channels/audioStream'
    summary: Caller DTMF key press
    description: >-
      Sent by Plivo when a DTMF digit is detected on the live call.
    messages:
      - $ref: '#/channels/audioStream/messages/dtmf'
  receivePlayedStream:
    action: receive
    channel:
      $ref: '#/channels/audioStream'
    summary: Checkpoint reached
    description: >-
      Sent by Plivo after audio queued by a prior `playAudio` event has played
      through the checkpoint identified by `name`. Allows the server to
      synchronize follow-up actions with playback completion.
    messages:
      - $ref: '#/channels/audioStream/messages/playedStream'
  receiveClearedAudio:
    action: receive
    channel:
      $ref: '#/channels/audioStream'
    summary: Buffered audio cleared
    description: >-
      Sent by Plivo to acknowledge that buffered playback audio has been
      cleared in response to a server-sent `clearAudio` event.
    messages:
      - $ref: '#/channels/audioStream/messages/clearedAudio'
  sendPlayAudio:
    action: send
    channel:
      $ref: '#/channels/audioStream'
    summary: Play audio into the call
    description: >-
      Sent by the server (only when `bidirectional="true"` on the `<Stream>`
      XML) to inject base64-encoded audio into the live call. `contentType`
      and `sampleRate` must match the stream's negotiated media format.
    messages:
      - $ref: '#/channels/audioStream/messages/playAudio'
  sendCheckpoint:
    action: send
    channel:
      $ref: '#/channels/audioStream'
    summary: Mark a playback checkpoint
    description: >-
      Sent by the server to label a position in the outbound playback queue.
      Plivo responds with a `playedStream` event when audio queued before the
      checkpoint has finished playing.
    messages:
      - $ref: '#/channels/audioStream/messages/checkpoint'
  sendClearAudio:
    action: send
    channel:
      $ref: '#/channels/audioStream'
    summary: Clear buffered playback audio
    description: >-
      Sent by the server to interrupt and discard any buffered playback audio
      previously sent via `playAudio`. Plivo responds with `clearedAudio`.
    messages:
      - $ref: '#/channels/audioStream/messages/clearAudio'
  sendDTMF:
    action: send
    channel:
      $ref: '#/channels/audioStream'
    summary: Send DTMF digits into the call
    description: >-
      Sent by the server (when bidirectional streaming is active) to play DTMF
      digits into the live call.
    messages:
      - $ref: '#/channels/audioStream/messages/sendDTMF'

components:
  messages:
    StartEvent:
      name: start
      title: Stream Start
      summary: Initial stream metadata sent by Plivo on connection.
      contentType: application/json
      payload:
        $ref: '#/components/schemas/StartPayload'
    MediaEvent:
      name: media
      title: Media Chunk
      summary: Base64-encoded raw audio chunk from the live call.
      contentType: application/json
      payload:
        $ref: '#/components/schemas/MediaPayload'
    DtmfEvent:
      name: dtmf
      title: DTMF Digit
      summary: A DTMF digit detected on the live call.
      contentType: application/json
      payload:
        $ref: '#/components/schemas/DtmfPayload'
    PlayedStreamEvent:
      name: playedStream
      title: Checkpoint Played
      summary: Acknowledgement that playback has reached a named checkpoint.
      contentType: application/json
      payload:
        $ref: '#/components/schemas/PlayedStreamPayload'
    ClearedAudioEvent:
      name: clearedAudio
      title: Buffered Audio Cleared
      summary: Acknowledgement that buffered playback audio was cleared.
      contentType: application/json
      payload:
        $ref: '#/components/schemas/ClearedAudioPayload'
    PlayAudioEvent:
      name: playAudio
      title: Play Audio
      summary: Server-to-Plivo audio injection during bidirectional streaming.
      contentType: application/json
      payload:
        $ref: '#/components/schemas/PlayAudioPayload'
    CheckpointEvent:
      name: checkpoint
      title: Checkpoint
      summary: Server-to-Plivo playback checkpoint marker.
      contentType: application/json
      payload:
        $ref: '#/components/schemas/CheckpointPayload'
    ClearAudioEvent:
      name: clearAudio
      title: Clear Audio
      summary: Server-to-Plivo request to discard buffered playback audio.
      contentType: application/json
      payload:
        $ref: '#/components/schemas/ClearAudioPayload'
    SendDTMFEvent:
      name: sendDTMF
      title: Send DTMF
      summary: Server-to-Plivo request to play DTMF digits into the call.
      contentType: application/json
      payload:
        $ref: '#/components/schemas/SendDTMFPayload'

  schemas:
    StartPayload:
      type: object
      required:
        - event
        - sequenceNumber
        - start
      properties:
        event:
          type: string
          const: start
          description: Event discriminator.
        sequenceNumber:
          type: integer
          minimum: 1
          description: Monotonically increasing per-stream sequence number.
        start:
          type: object
          required:
            - callId
            - streamId
            - accountId
            - tracks
            - mediaFormat
          properties:
            callId:
              type: string
              format: uuid
              description: Plivo call UUID for the live call being streamed.
            streamId:
              type: string
              format: uuid
              description: Unique Plivo identifier for this audio stream.
            accountId:
              type: string
              description: Plivo account (Auth ID) under which the call is running.
            tracks:
              type: array
              description: >-
                Audio tracks included in this stream, as configured by the
                `audioTrack` attribute on the `<Stream>` XML element.
              items:
                type: string
                enum:
                  - inbound
                  - outbound
            mediaFormat:
              type: object
              required:
                - encoding
                - sampleRate
              properties:
                encoding:
                  type: string
                  description: Audio encoding MIME type used on the wire.
                  enum:
                    - audio/x-l16
                    - audio/x-mulaw
                sampleRate:
                  type: integer
                  description: Sample rate in Hertz.
                  enum:
                    - 8000
                    - 16000
        extra_headers:
          type: string
          description: >-
            Custom key-value pairs forwarded from the `extraHeaders` attribute
            of the originating `<Stream>` XML element.

    MediaPayload:
      type: object
      required:
        - event
        - sequenceNumber
        - streamId
        - media
      properties:
        event:
          type: string
          const: media
        sequenceNumber:
          type: integer
          description: Per-stream sequence number.
        streamId:
          type: string
          format: uuid
          description: Plivo stream identifier.
        media:
          type: object
          required:
            - track
            - chunk
            - timestamp
            - payload
          properties:
            track:
              type: string
              description: The audio track this chunk belongs to.
              enum:
                - inbound
                - outbound
            chunk:
              type: integer
              description: Sequence number of this chunk within the stream.
            timestamp:
              type: string
              description: Unix epoch timestamp in milliseconds, as a string.
            payload:
              type: string
              format: byte
              description: >-
                Base64-encoded raw audio payload (approximately 20ms of audio
                per chunk).
        extra_headers:
          type: string
          description: Custom headers forwarded from the originating `<Stream>` XML.

    DtmfPayload:
      type: object
      required:
        - event
        - sequenceNumber
        - streamId
        - dtmf
      properties:
        event:
          type: string
          const: dtmf
        sequenceNumber:
          type: integer
        streamId:
          type: string
          format: uuid
        dtmf:
          type: object
          required:
            - track
            - digit
            - timestamp
          properties:
            track:
              type: string
              enum:
                - inbound
                - outbound
            digit:
              type: string
              description: A single DTMF digit.
              enum:
                - '0'
                - '1'
                - '2'
                - '3'
                - '4'
                - '5'
                - '6'
                - '7'
                - '8'
                - '9'
                - '*'
                - '#'
                - 'A'
                - 'B'
                - 'C'
                - 'D'
            timestamp:
              type: string
              description: Unix epoch timestamp in milliseconds, as a string.
        extra_headers:
          type: string

    PlayedStreamPayload:
      type: object
      required:
        - event
        - sequenceNumber
        - streamId
        - name
      properties:
        event:
          type: string
          const: playedStream
        sequenceNumber:
          type: integer
        streamId:
          type: string
          format: uuid
        name:
          type: string
          description: >-
            Identifier of the checkpoint previously declared by the server
            with a `checkpoint` event. Emitted by Plivo when playback has
            advanced through that checkpoint.

    ClearedAudioPayload:
      type: object
      required:
        - event
        - sequenceNumber
        - streamId
      properties:
        event:
          type: string
          const: clearedAudio
        sequenceNumber:
          type: integer
        streamId:
          type: string
          format: uuid

    PlayAudioPayload:
      type: object
      required:
        - event
        - media
      properties:
        event:
          type: string
          const: playAudio
        media:
          type: object
          required:
            - contentType
            - sampleRate
            - payload
          properties:
            contentType:
              type: string
              description: >-
                MIME type of the supplied audio. Must match the stream's
                negotiated encoding.
              enum:
                - audio/x-l16
                - audio/x-mulaw
            sampleRate:
              description: >-
                Sample rate of the supplied audio in Hertz. Must match the
                stream's negotiated sample rate. Plivo accepts this field as
                either a number or a numeric string.
              oneOf:
                - type: integer
                  enum:
                    - 8000
                    - 16000
                - type: string
                  enum:
                    - '8000'
                    - '16000'
            payload:
              type: string
              format: byte
              description: Base64-encoded raw audio payload to inject into the call.

    CheckpointPayload:
      type: object
      required:
        - event
        - streamId
        - name
      properties:
        event:
          type: string
          const: checkpoint
        streamId:
          type: string
          format: uuid
        name:
          type: string
          description: >-
            Unique server-chosen checkpoint identifier. Plivo will echo this
            back in a `playedStream` event once buffered playback has reached
            this point.

    ClearAudioPayload:
      type: object
      required:
        - event
        - streamId
      properties:
        event:
          type: string
          const: clearAudio
        streamId:
          type: string
          format: uuid

    SendDTMFPayload:
      type: object
      required:
        - event
        - dtmf
      properties:
        event:
          type: string
          const: sendDTMF
        dtmf:
          type: string
          description: One or more DTMF digits to play into the live call.
          pattern: '^[0-9A-D*#]+$'