> ## Documentation Index > Fetch the complete documentation index at: https://developers.telnyx.com/llms.txt > Use this file to discover all available pages before exploring further. # Remove background noise over WebSocket > Stream raw mono PCM audio and receive denoised audio in the same format and sample rate. Set all options as query parameters on the upgrade request. They cannot change during the session. A request without valid credentials fails with HTTP `401`. A `sample_rate` or `channels` value that is not a valid non-negative integer fails with HTTP `400`. While the server restarts, new connections fail with HTTP `503`; reconnect with backoff. Any other unsupported parameter value closes the connection with an `invalid_params` error frame and close code `1008`. Sessions are limited to 590 seconds. At the limit, the server returns the remaining audio, sends a `max_session_duration` error frame, and closes with `1008`. ## AsyncAPI ````yaml openapi/asyncapi/noise-suppression-websockets/noise-suppression-websockets.yml noise-suppression-websockets id: noise-suppression-websockets title: Remove background noise over WebSocket description: > Stream raw mono PCM audio and receive denoised audio in the same format and sample rate. Set all options as query parameters on the upgrade request. They cannot change during the session. A request without valid credentials fails with HTTP `401`. A `sample_rate` or `channels` value that is not a valid non-negative integer fails with HTTP `400`. While the server restarts, new connections fail with HTTP `503`; reconnect with backoff. Any other unsupported parameter value closes the connection with an `invalid_params` error frame and close code `1008`. Sessions are limited to 590 seconds. At the limit, the server returns the remaining audio, sends a `max_session_duration` error frame, and closes with `1008`. servers: - id: noise-suppression-websockets protocol: wss host: api.telnyx.com bindings: - protocol: ws version: 0.1.0 value: {} schemaProperties: [] variables: [] address: /v2/noise_suppression parameters: [] bindings: - protocol: ws version: 0.1.0 value: query: type: object description: Query parameters passed when opening the WebSocket connection. properties: format: type: string description: > Audio encoding, used for both input and output. `l16` is 16-bit signed integer PCM, little-endian (2 bytes per sample). `f32le` is 32-bit float PCM in the range -1.0 to 1.0, little-endian (4 bytes per sample). enum: - l16 - f32le default: l16 sample_rate: type: integer description: > Sample rate in Hz, used for both input and output. Any integer from 8000 to 48000. A value outside that range closes the connection with `invalid_params` and close code `1008`. Send telephony and PSTN audio at `8000`. Do not upsample 8 kHz audio to a higher rate: the full-band model treats the missing upper band as noise and lowers the speech level by about 14 dB. minimum: 8000 maximum: 48000 default: 16000 channels: type: integer description: Number of channels. Only mono audio is supported. enum: - 1 default: 1 model: type: string description: > Noise suppression model. The service selects the model variant from `sample_rate`. `rook` is for audio that people listen to: the 8 kHz model at 8000, the 16 kHz model from 8001 to 16000, and the 48 kHz model above 16000. `quail` is for audio that goes to speech-to-text: the 8 kHz model at 8000 and the 16 kHz model above 8000, so its output bandwidth is limited to 8 kHz. enum: - rook - quail default: rook quality: description: > Enhancement strength: how much noise the model removes. A number from `0.0` (lightest processing) to `1.0` (strongest), or a preset: `low_latency` (0.5, lighter processing; the name is kept for compatibility), `balanced` (0.8), or `high_quality` (1.0). default: balanced oneOf: - type: number minimum: 0 maximum: 1 - type: string enum: - low_latency - balanced - high_quality schemaProperties: - name: query type: object description: Query parameters passed when opening the WebSocket connection. required: false properties: - name: format type: string description: > Audio encoding, used for both input and output. `l16` is 16-bit signed integer PCM, little-endian (2 bytes per sample). `f32le` is 32-bit float PCM in the range -1.0 to 1.0, little-endian (4 bytes per sample). enumValues: - l16 - f32le required: false - name: sample_rate type: integer description: > Sample rate in Hz, used for both input and output. Any integer from 8000 to 48000. A value outside that range closes the connection with `invalid_params` and close code `1008`. Send telephony and PSTN audio at `8000`. Do not upsample 8 kHz audio to a higher rate: the full-band model treats the missing upper band as noise and lowers the speech level by about 14 dB. required: false - name: channels type: integer description: Number of channels. Only mono audio is supported. enumValues: - 1 required: false - name: model type: string description: > Noise suppression model. The service selects the model variant from `sample_rate`. `rook` is for audio that people listen to: the 8 kHz model at 8000, the 16 kHz model from 8001 to 16000, and the 48 kHz model above 16000. `quail` is for audio that goes to speech-to-text: the 8 kHz model at 8000 and the 16 kHz model above 8000, so its output bandwidth is limited to 8 kHz. enumValues: - rook - quail required: false - name: quality type: oneOf description: > Enhancement strength: how much noise the model removes. A number from `0.0` (lightest processing) to `1.0` (strongest), or a preset: `low_latency` (0.5, lighter processing; the name is kept for compatibility), `balanced` (0.8), or `high_quality` (1.0). required: false operations: - &ref_1 id: sendAudio title: Send Audio to Denoise description: >- Client sends binary audio frames, then an end_of_stream message when it has no more audio. type: receive messages: - &ref_3 id: clientAudio contentType: application/octet-stream payload: - type: string format: binary description: > Client-to-server binary frame with raw mono PCM audio, in the `format` and `sample_rate` set on the connection. No container and no WAV header. The frame must contain a whole number of samples (a multiple of 2 bytes for `l16`, 4 bytes for `f32le`) and must not exceed 65,536 bytes. 20 ms chunks are a good default. x-parser-schema-id: name: Audio Frame headers: [] jsonPayloadSchema: type: string format: binary description: Raw PCM audio samples, little-endian. x-parser-schema-id: title: Audio Frame description: > Client-to-server binary frame with raw mono PCM audio, in the `format` and `sample_rate` set on the connection. No container and no WAV header. The frame must contain a whole number of samples (a multiple of 2 bytes for `l16`, 4 bytes for `f32le`) and must not exceed 65,536 bytes. 20 ms chunks are a good default. example: '{}' bindings: [] extensions: - id: x-parser-unique-object-id value: clientAudio - &ref_4 id: clientEndOfStream contentType: application/json payload: - name: End of Stream description: > Client-to-server text frame. Send it when you have no more audio. The server returns all remaining audio, including the audio held back by the model's processing delay, as one final binary frame, then closes with code `1000`. If you close the connection without it, the end of your stream is lost. Any other text frame gets an `unsupported_message` error frame and the session continues. type: object properties: - name: type type: string description: Message type identifier. required: true headers: [] jsonPayloadSchema: type: object required: - type properties: type: type: string const: end_of_stream description: Message type identifier. x-parser-schema-id: x-parser-schema-id: title: End of Stream description: > Client-to-server text frame. Send it when you have no more audio. The server returns all remaining audio, including the audio held back by the model's processing delay, as one final binary frame, then closes with code `1000`. If you close the connection without it, the end of your stream is lost. Any other text frame gets an `unsupported_message` error frame and the session continues. example: |- { "type": "end_of_stream" } bindings: [] extensions: - id: x-parser-unique-object-id value: clientEndOfStream bindings: [] extensions: &ref_0 - id: x-parser-unique-object-id value: noise-suppression-websockets - &ref_2 id: receiveAudio title: Receive Denoised Audio description: Server sends denoised audio frames and error frames to the client. type: send messages: - &ref_5 id: serverAudio contentType: application/octet-stream payload: - type: string format: binary description: > Server-to-client binary frame with denoised audio, in the same `format` and `sample_rate` as the input. The model processes fixed-size internal frames and has a short processing delay, so a frame can be smaller or larger than the matching input frame, or empty. Ignore empty frames. Over the session, total output equals the input length plus the processing delay. x-parser-schema-id: name: Denoised Audio Frame headers: [] jsonPayloadSchema: type: string format: binary description: Denoised raw PCM audio samples, little-endian. x-parser-schema-id: title: Denoised Audio Frame description: > Server-to-client binary frame with denoised audio, in the same `format` and `sample_rate` as the input. The model processes fixed-size internal frames and has a short processing delay, so a frame can be smaller or larger than the matching input frame, or empty. Ignore empty frames. Over the session, total output equals the input length plus the processing delay. example: '{}' bindings: [] extensions: - id: x-parser-unique-object-id value: serverAudio - &ref_6 id: serverError contentType: application/json payload: - name: Error Frame description: > Server-to-client text frame. If the error ends the session, the server closes the connection right after this frame, with the close code listed below and the error code as the close reason. - `invalid_params` (close `1008`): a query parameter has an unsupported value. - `max_session_duration` (close `1008`): the session reached the 590-second limit. The remaining audio is returned first. - `server_shutdown` (close `1001`): the server is restarting. The remaining audio is returned first. Reconnect with backoff. - `session_error` (close `1011`): the server could not start the noise suppression session. - `processing_error` (close `1011`): the noise suppression engine failed. - `encode_error` (close `1011`): the server could not encode the denoised audio. - `max_connections` (close `1013`): the server has too many open sessions. Try again later. - `overloaded` (close `1013`): the server is overloaded and does not accept new sessions. Try again later. - `chunk_too_large` (session continues): the binary frame is larger than 65,536 bytes. The chunk is dropped. - `decode_error` (session continues): the frame length is not a whole number of samples. The chunk is dropped. - `unsupported_message` (session continues): the text frame is not a supported message. type: object properties: - name: type type: string description: Frame type identifier. required: true - name: code type: string description: Stable error code. enumValues: - invalid_params - max_session_duration - server_shutdown - session_error - processing_error - encode_error - max_connections - overloaded - chunk_too_large - decode_error - unsupported_message required: true - name: message type: string description: Human-readable description. Do not parse it. required: true headers: [] jsonPayloadSchema: type: object required: - type - code - message properties: type: type: string const: error description: Frame type identifier. x-parser-schema-id: code: type: string description: Stable error code. enum: - invalid_params - max_session_duration - server_shutdown - session_error - processing_error - encode_error - max_connections - overloaded - chunk_too_large - decode_error - unsupported_message x-parser-schema-id: message: type: string description: Human-readable description. Do not parse it. x-parser-schema-id: x-parser-schema-id: title: Error Frame description: > Server-to-client text frame. If the error ends the session, the server closes the connection right after this frame, with the close code listed below and the error code as the close reason. - `invalid_params` (close `1008`): a query parameter has an unsupported value. - `max_session_duration` (close `1008`): the session reached the 590-second limit. The remaining audio is returned first. - `server_shutdown` (close `1001`): the server is restarting. The remaining audio is returned first. Reconnect with backoff. - `session_error` (close `1011`): the server could not start the noise suppression session. - `processing_error` (close `1011`): the noise suppression engine failed. - `encode_error` (close `1011`): the server could not encode the denoised audio. - `max_connections` (close `1013`): the server has too many open sessions. Try again later. - `overloaded` (close `1013`): the server is overloaded and does not accept new sessions. Try again later. - `chunk_too_large` (session continues): the binary frame is larger than 65,536 bytes. The chunk is dropped. - `decode_error` (session continues): the frame length is not a whole number of samples. The chunk is dropped. - `unsupported_message` (session continues): the text frame is not a supported message. example: |- { "type": "error", "code": "invalid_params", "message": "unsupported sample_rate 96000 (supported: 8000-48000 Hz)" } bindings: [] extensions: - id: x-parser-unique-object-id value: serverError bindings: [] extensions: *ref_0 sendOperations: - *ref_1 receiveOperations: - *ref_2 sendMessages: - *ref_3 - *ref_4 receiveMessages: - *ref_5 - *ref_6 extensions: - id: x-parser-unique-object-id value: noise-suppression-websockets securitySchemes: - id: bearerAuth name: bearerAuth type: http description: Telnyx API v2 Bearer token authentication. scheme: bearer extensions: [] ````