> ## Documentation Index > Fetch the complete documentation index at: https://docs.slng.ai/llms.txt > Use this file to discover all available pages before exploring further. # Nova 3 > Stream real-time speech-to-text transcripts from Deepgram Nova 3 over WebSocket with voice activity detection, speaker diarization, and partial results. **Hosting: routed through SLNG (Deepgram).** This model runs on Deepgram's servers; SLNG proxies each request through the nearest regional gateway. Available in Netherlands, Germany, United States (East, West), Australia, Brazil, United Kingdom, Indonesia, Israel, India, Japan, Singapore, South Africa. Connect to your region's gateway WebSocket endpoint: `wss://eu-north.api.slng.ai`, `wss://eu-west.api.slng.ai`, `wss://us-east.api.slng.ai`, `wss://us-west.api.slng.ai`, `wss://au.api.slng.ai`, `wss://br.api.slng.ai`, `wss://gb.api.slng.ai`, `wss://id.api.slng.ai`, `wss://il.api.slng.ai`, `wss://in.api.slng.ai`, `wss://jp.api.slng.ai`, `wss://sg.api.slng.ai`, `wss://za.api.slng.ai`. ## AsyncAPI ````yaml api-reference/asyncapi/stt-deepgram.asyncapi.yaml /v1/stt/deepgram/nova:3 id: /v1/stt/deepgram/nova:3 title: Nova 3 description: >- Stream real-time speech-to-text transcripts from Deepgram Nova 3 over WebSocket with voice activity detection, speaker diarization, and partial results. servers: - id: production-eu-north protocol: wss host: eu-north.api.slng.ai bindings: [] variables: [] - id: production-eu-west protocol: wss host: eu-west.api.slng.ai bindings: [] variables: [] - id: production-us-east protocol: wss host: us-east.api.slng.ai bindings: [] variables: [] - id: production-us-west protocol: wss host: us-west.api.slng.ai bindings: [] variables: [] - id: production-au protocol: wss host: au.api.slng.ai bindings: [] variables: [] - id: production-br protocol: wss host: br.api.slng.ai bindings: [] variables: [] - id: production-gb protocol: wss host: gb.api.slng.ai bindings: [] variables: [] - id: production-id protocol: wss host: id.api.slng.ai bindings: [] variables: [] - id: production-il protocol: wss host: il.api.slng.ai bindings: [] variables: [] - id: production-in protocol: wss host: in.api.slng.ai bindings: [] variables: [] - id: production-jp protocol: wss host: jp.api.slng.ai bindings: [] variables: [] - id: production-sg protocol: wss host: sg.api.slng.ai bindings: [] variables: [] - id: production-za protocol: wss host: za.api.slng.ai bindings: [] variables: [] address: /v1/stt/deepgram/nova:3 parameters: [] bindings: - protocol: ws version: latest value: method: GET schemaProperties: - name: method type: string description: GET required: false operations: - &ref_1 id: sttDeepgramNova3ReceiveInit title: Stt deepgram nova3 receive init description: Initialize session on Nova 3 type: receive messages: - &ref_9 id: DeepgramSttInitRequest contentType: application/json payload: - name: Init Request description: >- Initialize a Deepgram STT session with provider-specific recognition configuration. type: object properties: - name: type type: string description: init required: true - name: config type: object description: Deepgram recognition configuration options required: false properties: - name: language type: string description: Supported language code for transcription. enumValues: - multi - en - en-us - en-au - en-gb - en-nz - en-in - en-ca - es - fr - de - it - pt - nl - hi - ja - ru required: false - name: sample_rate type: integer description: Input audio sample rate in Hz. required: false - name: encoding type: string description: Input audio encoding format. enumValues: - linear16 - flac - mulaw - amr-nb - amr-wb - opus - speex - mp3 - mp4 - webm - aac - ogg required: false - name: enable_partials type: boolean description: Enable partial/interim transcription results. required: false - name: punctuate type: boolean description: Add punctuation and capitalization. required: false - name: smart_format type: boolean description: >- Apply formatting to improve readability (dates, times, numbers, etc.). required: false - name: diarize type: boolean description: >- Enable speaker diarization. Words assigned speaker numbers starting at 0. required: false - name: numerals type: boolean description: Convert spoken numbers to digits. required: false - name: profanity_filter type: boolean description: Filter profanity from transcript. required: false - name: endpointing type: integer description: >- Time in milliseconds of silence before finalizing speech. Set to 0 to disable endpointing entirely. required: false - name: utterance_end_ms type: integer description: >- Time in milliseconds of silence between words before sending an UtteranceEnd message. Default is 1000ms, minimum 200ms, maximum 5000ms. Requires enable_partials to be true. required: false headers: [] jsonPayloadSchema: type: object description: >- Initialize a Deepgram Nova STT session with full provider-specific configuration options before streaming audio. required: - type properties: type: type: string const: init x-parser-schema-id: config: type: object description: Deepgram recognition configuration options properties: language: type: string description: Supported language code for transcription. default: en enum: - multi - en - en-us - en-au - en-gb - en-nz - en-in - en-ca - es - fr - de - it - pt - nl - hi - ja - ru x-parser-schema-id: sample_rate: type: integer description: Input audio sample rate in Hz. default: 16000 x-parser-schema-id: encoding: type: string description: Input audio encoding format. default: linear16 enum: - linear16 - flac - mulaw - amr-nb - amr-wb - opus - speex - mp3 - mp4 - webm - aac - ogg x-parser-schema-id: enable_partials: type: boolean description: Enable partial/interim transcription results. default: true x-parser-schema-id: punctuate: type: boolean description: Add punctuation and capitalization. default: false x-parser-schema-id: smart_format: type: boolean description: >- Apply formatting to improve readability (dates, times, numbers, etc.). default: false x-parser-schema-id: diarize: type: boolean description: >- Enable speaker diarization. Words assigned speaker numbers starting at 0. default: false x-parser-schema-id: numerals: type: boolean description: Convert spoken numbers to digits. default: false x-parser-schema-id: profanity_filter: type: boolean description: Filter profanity from transcript. default: false x-parser-schema-id: endpointing: type: integer minimum: 0 default: 10 description: >- Time in milliseconds of silence before finalizing speech. Set to 0 to disable endpointing entirely. x-parser-schema-id: utterance_end_ms: type: integer minimum: 200 maximum: 5000 default: 1000 description: >- Time in milliseconds of silence between words before sending an UtteranceEnd message. Default is 1000ms, minimum 200ms, maximum 5000ms. Requires enable_partials to be true. x-parser-schema-id: x-parser-schema-id: x-parser-schema-id: title: Init Request description: >- Initialize a Deepgram STT session with provider-specific recognition configuration. example: |- { "type": "init", "config": { "language": "en", "sample_rate": 16000, "encoding": "linear16", "enable_partials": true, "punctuate": true, "smart_format": true } } bindings: [] extensions: - id: x-parser-unique-object-id value: DeepgramSttInitRequest bindings: [] extensions: &ref_0 - id: x-parser-unique-object-id value: /v1/stt/deepgram/nova:3 - &ref_2 id: sttDeepgramNova3ReceiveAudio title: Stt deepgram nova3 receive audio description: Send audio frame to Nova 3 type: receive messages: - &ref_10 id: DeepgramSttBinaryAudio contentType: application/octet-stream payload: - type: string format: binary description: > Stream raw binary PCM audio frames to be transcribed. Sent as binary WebSocket frames (NOT JSON). Recommended chunk size is 8192 bytes (~256ms at 16kHz linear16). Start streaming immediately after the init message — do not wait for the Metadata response. x-parser-schema-id: name: Binary Audio Frame headers: [] jsonPayloadSchema: type: string format: binary description: > Raw binary audio data (linear16 PCM at 16kHz mono). Sent as a binary WebSocket frame — no JSON envelope, no base64 encoding. Each frame is a chunk of raw PCM samples. Example hex dump of one frame (speech waveform): 0000FF00000000FF00000000010101010101010100000000FFFFFFFEFE FEFEFDFEFEFEFEFDFDFEFEFEFEFEFEFEFEFEFFFFFFFFFEFEFEFEFF000100 001020303030303030303030201010000FFFFFEFDFDFDFDFEFFFFFFFF00 Each pair of hex digits is one byte. At 16kHz linear16, 8192 bytes = 4096 samples = ~256ms of audio. x-parser-schema-id: title: Binary Audio Frame description: > Stream raw binary PCM audio frames to be transcribed. Sent as binary WebSocket frames (NOT JSON). Recommended chunk size is 8192 bytes (~256ms at 16kHz linear16). Start streaming immediately after the init message — do not wait for the Metadata response. example: >- "0000FF00000000FF00000000010101010101010100000000FFFFFFFFFFFFFEFEFDFEFEFEFEFDFDFEFEFEFEFEFEFEFEFEFFFFFFFFFEFEFEFEFF000100001020303030303030303030201010000FFFFFEFDFDFDFDFEFFFFFFFF0001020303020201000000FFFDFCFBFAFAFBFAF9F8F7F7F7F6F6F4F2F2F3F7FC000406090F14191A19181715110E0A05FEF9F6F3F0EEECEBEBECEEF2F6F9FC0005090D0F101010100E0C080401" bindings: [] extensions: - id: x-parser-unique-object-id value: DeepgramSttBinaryAudio bindings: [] extensions: *ref_0 - &ref_3 id: sttDeepgramNova3ReceiveFinalize title: Stt deepgram nova3 receive finalize description: Force-finalize buffered audio on Nova 3 type: receive messages: - &ref_11 id: SttFinalizeMessage contentType: application/json payload: - name: Finalize Message description: >- Force-finalize buffered audio tokens without closing the connection. type: object properties: - name: type type: string description: finalize required: true headers: [] jsonPayloadSchema: type: object description: >- Mid-stream flush — forces the server to finalize any buffered audio and return results. The connection remains open for further audio. required: - type properties: type: type: string const: finalize x-parser-schema-id: x-parser-schema-id: title: Finalize Message description: Force-finalize buffered audio tokens without closing the connection. example: |- { "type": "finalize" } bindings: [] extensions: - id: x-parser-unique-object-id value: SttFinalizeMessage bindings: [] extensions: *ref_0 - &ref_4 id: sttDeepgramNova3ReceiveClose title: Stt deepgram nova3 receive close description: Close audio stream on Nova 3 type: receive messages: - &ref_12 id: SttCloseMessage contentType: application/json payload: - name: Close Message description: Signal end of audio stream and close the connection. type: object properties: - name: type type: string description: close required: true headers: [] jsonPayloadSchema: type: object description: >- Signal that no more audio will be sent. The server processes remaining audio, sends final results, then closes the connection. required: - type properties: type: type: string const: close x-parser-schema-id: x-parser-schema-id: title: Close Message description: Signal end of audio stream and close the connection. example: |- { "type": "close" } bindings: [] extensions: - id: x-parser-unique-object-id value: SttCloseMessage bindings: [] extensions: *ref_0 - &ref_5 id: sttDeepgramNova3ReceiveKeepalive title: Stt deepgram nova3 receive keepalive description: Send keepalive on Nova 3 type: receive messages: - &ref_13 id: KeepaliveMessage contentType: application/json payload: - name: Keepalive Message description: Keep the WebSocket connection alive during silence. type: object properties: - name: type type: string description: keepalive required: true headers: [] jsonPayloadSchema: type: object description: >- Prevents the connection from being closed due to inactivity. Send periodically during silence to maintain the session. required: - type properties: type: type: string const: keepalive x-parser-schema-id: x-parser-schema-id: title: Keepalive Message description: Keep the WebSocket connection alive during silence. example: |- { "type": "keepalive" } bindings: [] extensions: - id: x-parser-unique-object-id value: KeepaliveMessage bindings: [] extensions: *ref_0 - &ref_6 id: sttDeepgramNova3SendMetadata title: Stt deepgram nova3 send metadata description: Receive Deepgram session metadata from Nova 3 type: send messages: - &ref_14 id: DeepgramSttMetadata contentType: application/json payload: - name: Metadata description: > Deepgram session metadata. Sent when the backend connects and again after the session ends with final statistics. This is informational — do not wait for it before streaming audio. type: object properties: - name: type type: string description: Metadata required: true - name: request_id type: string description: Unique Deepgram request identifier required: false - name: created type: string description: Timestamp when the session was created required: false - name: duration type: number description: >- Total audio duration processed in seconds (0 on initial, populated on final) required: false - name: channels type: integer description: Number of audio channels required: false - name: sha256 type: string description: SHA-256 hash of the audio (populated on final Metadata only) required: false - name: transaction_key type: string description: Deprecated field (always "deprecated") required: false headers: [] jsonPayloadSchema: type: object description: > Deepgram native Metadata message. Received twice: once when the session opens (duration=0) and once after finalize with the full session summary. required: - type properties: type: type: string const: Metadata x-parser-schema-id: request_id: type: string description: Unique Deepgram request identifier x-parser-schema-id: created: type: string format: date-time description: Timestamp when the session was created x-parser-schema-id: duration: type: number description: >- Total audio duration processed in seconds (0 on initial, populated on final) x-parser-schema-id: channels: type: integer description: Number of audio channels x-parser-schema-id: sha256: type: string description: SHA-256 hash of the audio (populated on final Metadata only) x-parser-schema-id: transaction_key: type: string description: Deprecated field (always "deprecated") x-parser-schema-id: x-parser-schema-id: title: Metadata description: > Deepgram session metadata. Sent when the backend connects and again after the session ends with final statistics. This is informational — do not wait for it before streaming audio. example: |- { "type": "Metadata", "request_id": "a2b1db35-40ff-41f9-9912-dedaeab02228", "created": "2026-04-01T09:40:09.152Z", "duration": 0, "channels": 1, "sha256": "incomplete", "transaction_key": "deprecated" } bindings: [] extensions: - id: x-parser-unique-object-id value: DeepgramSttMetadata bindings: [] extensions: *ref_0 - &ref_7 id: sttDeepgramNova3SendResults title: Stt deepgram nova3 send results description: Receive transcription results from Deepgram Nova 3 type: send messages: - &ref_15 id: DeepgramSttResultsEn contentType: application/json payload: - name: Results description: >- Deepgram transcription result — partial or final, determined by is_final. type: object properties: - name: type type: string description: Results required: true - name: is_final type: boolean description: true for final results, false for interim/partial required: false - name: speech_final type: boolean description: true when endpoint detection fires (end of utterance) required: false - name: channel_index type: array description: Channel index pair [channel, total_channels] required: false properties: - name: item type: integer required: false - name: channel type: object description: Deepgram channel result with transcription alternatives required: true properties: - name: alternatives type: array required: false properties: - name: transcript type: string description: Transcribed text required: false - name: confidence type: number description: Confidence score (0-1) required: false - name: words type: array description: Word-level transcription details required: false properties: - name: word type: string required: false - name: start type: number description: Word start time in seconds required: false - name: end type: number description: Word end time in seconds required: false - name: confidence type: number required: false - name: speaker type: integer description: Speaker ID (when diarization enabled) required: false - name: punctuated_word type: string required: false - name: start type: number description: Start time of this segment in seconds required: false - name: duration type: number description: Duration of this segment in seconds required: false headers: [] jsonPayloadSchema: type: object description: > Deepgram native Results message. Contains the transcription in channel.alternatives[0].transcript. Use is_final to distinguish partial (interim) from final results, and speech_final to detect end-of-utterance boundaries. required: - type - channel properties: type: type: string const: Results x-parser-schema-id: is_final: type: boolean description: true for final results, false for interim/partial x-parser-schema-id: speech_final: type: boolean description: true when endpoint detection fires (end of utterance) x-parser-schema-id: channel_index: type: array items: type: integer x-parser-schema-id: description: Channel index pair [channel, total_channels] x-parser-schema-id: channel: type: object description: Deepgram channel result with transcription alternatives properties: alternatives: type: array items: type: object properties: transcript: type: string description: Transcribed text x-parser-schema-id: confidence: type: number description: Confidence score (0-1) x-parser-schema-id: words: type: array description: Word-level transcription details items: type: object properties: word: type: string x-parser-schema-id: start: type: number description: Word start time in seconds x-parser-schema-id: end: type: number description: Word end time in seconds x-parser-schema-id: confidence: type: number x-parser-schema-id: speaker: type: integer description: Speaker ID (when diarization enabled) x-parser-schema-id: punctuated_word: type: string x-parser-schema-id: x-parser-schema-id: x-parser-schema-id: x-parser-schema-id: x-parser-schema-id: x-parser-schema-id: start: type: number description: Start time of this segment in seconds x-parser-schema-id: duration: type: number description: Duration of this segment in seconds x-parser-schema-id: x-parser-schema-id: DeepgramSttResultsPayload title: Results description: >- Deepgram transcription result — partial or final, determined by is_final. example: |- { "type": "Results", "channel_index": [ 0, 1 ], "duration": 3.52, "start": 0, "is_final": true, "speech_final": true, "channel": { "alternatives": [ { "transcript": "The quick brown fox jumps over the lazy dog.", "confidence": 0.9980469, "words": [ { "word": "the", "start": 0.08, "end": 0.16, "confidence": 0.9975586, "punctuated_word": "The" }, { "word": "quick", "start": 0.16, "end": 0.4, "confidence": 0.9951172, "punctuated_word": "quick" }, { "word": "brown", "start": 0.4, "end": 0.64, "confidence": 0.9980469, "punctuated_word": "brown" } ] } ] } } bindings: [] extensions: - id: x-parser-unique-object-id value: DeepgramSttResultsEn bindings: [] extensions: *ref_0 - &ref_8 id: sttDeepgramNova3SendError title: Stt deepgram nova3 send error description: Receive error from Nova 3 type: send messages: - &ref_16 id: DeepgramSttError contentType: application/json payload: - name: Error description: Error from the Deepgram backend or gateway. type: object properties: - name: type type: string description: Error required: true - name: code type: string description: >- Gateway error code (e.g., backend_connection_failed, stt_metering_unavailable) required: false - name: message type: string description: Human-readable error description (gateway errors) required: false - name: variant type: string description: Deepgram error variant (e.g., SchemaError) required: false - name: description type: string description: Deepgram error description required: false headers: [] jsonPayloadSchema: type: object description: > Error message. May originate from the gateway (with code field) or from Deepgram directly (with variant and description fields). required: - type properties: type: type: string const: Error x-parser-schema-id: code: type: string description: >- Gateway error code (e.g., backend_connection_failed, stt_metering_unavailable) x-parser-schema-id: message: type: string description: Human-readable error description (gateway errors) x-parser-schema-id: variant: type: string description: Deepgram error variant (e.g., SchemaError) x-parser-schema-id: description: type: string description: Deepgram error description x-parser-schema-id: x-parser-schema-id: title: Error description: Error from the Deepgram backend or gateway. example: |- { "type": "Error", "code": "backend_connection_failed", "message": "Failed to connect to backend: 500 Internal Server Error" } bindings: [] extensions: - id: x-parser-unique-object-id value: DeepgramSttError bindings: [] extensions: *ref_0 sendOperations: - *ref_1 - *ref_2 - *ref_3 - *ref_4 - *ref_5 receiveOperations: - *ref_6 - *ref_7 - *ref_8 sendMessages: - *ref_9 - *ref_10 - *ref_11 - *ref_12 - *ref_13 receiveMessages: - *ref_14 - *ref_15 - *ref_16 extensions: - id: x-parser-unique-object-id value: /v1/stt/deepgram/nova:3 securitySchemes: - id: bearer name: bearer type: http description: > API key issued by SLNG. Pass as `Authorization: Bearer ` in the WebSocket upgrade request headers. scheme: bearer extensions: [] ````