@audicle/sdk 0.1.1 → 0.1.2

package/README.md

@@ -1,6 +1,6 @@
 # Audicle Node.js SDK
 
-The official Node.js/TypeScript client for the [Audicle](https://audicle.ai) transcription API. Supports batch transcription, real-time streaming over WebSocket, and full transcript management.
+The official Node.js/TypeScript client for the [Audicle](https://audicle.ai) transcription API. Supports pre-recorded transcription, real-time streaming over WebSocket, and full transcript management.
 
 ## Installation
 
@@ -22,7 +22,7 @@ const result = await client.transcribe({
 console.log(result.result.text);
 ```
 
-## Batch Transcription
+## Pre-Recorded Transcription
 
 ### From a file
 
@@ -150,20 +150,9 @@ stream.sendAudio(pcmBuffer);
 stream.finalize();
 ```
 
-### Models
-
-Both models can be used for streaming via WebSocket. `audicle-o1-realtime` additionally supports interim (non-final) results and voice activity detection (VAD) events.
-
-| Model | Description |
-|---|---|
-| `audicle-o1` | Whisper-based — final results only |
-| `audicle-o1-realtime` | Low-latency — interim results + VAD |
-
 ### Observing a session
 
-Watch a live streaming session from another client. Only active WebSocket sessions can be observed — batch transcriptions submitted via file upload cannot.
-
-The `audicle-o1-realtime` model delivers interim results to observers in real time. The `audicle-o1` model only sends final results.
+Watch a live streaming session from another client. Only active WebSocket sessions can be observed — pre-recorded transcriptions submitted via file upload cannot.
 
 ```typescript
 const observer = client.streaming.observe("txn_abc123");

package/dist/index.d.cts

@@ -34,16 +34,12 @@ export interface paths {
 		};
 		/**
 		 * Stream audio for live transcription
-		 * @description Initiates a WebSocket connection for live audio transcription.
-		 *
-		 *     The `model` query parameter determines the transcription backend:
-		 *     - `audicle-o1` — Whisper-based streaming (TranscriptionSession DO)
-		 *     - `audicle-o1-realtime` — streaming provider relay (StreamingSession DO)
+		 * @description Initiates a WebSocket connection for live audio transcription. Only streaming models (e.g. `audicle-o1-realtime`) are supported — prerecorded models must use `POST /v1/transcribe`.
 		 *
 		 *     **Protocol:** Send binary audio frames (PCM s16le) over the WebSocket. The server responds with JSON messages:
 		 *     - `{"type": "session.begin", "id": "...", "model": "..."}` — connection established
 		 *     - `{"type": "transcript", "is_final": true/false, "transcript": {"text": "...", "start": 0.0, "end": 1.5}}` — transcription results
-		 *     - `{"type": "vad", "event": "speech_start|speech_end", "timestamp_ms": N}` — voice activity detection (streaming models only)
+		 *     - `{"type": "vad", "event": "speech_start|speech_end", "timestamp_ms": N}` — voice activity detection
 		 *     - `{"type": "error", "code": "...", "message": "..."}` — error occurred
 		 *     - `{"type": "session.end", "reason": "...", "usage": {...}}` — session ended
 		 *
@@ -240,6 +236,7 @@ export interface components {
 			end: number;
 			text: string;
 			confidence?: number;
+			speaker?: number;
 			words?: components["schemas"]["TranscriptionWord"][];
 		};
 		TranscriptionWord: {
@@ -247,6 +244,7 @@ export interface components {
 			start: number;
 			end: number;
 			confidence?: number;
+			speaker?: number;
 		};
 		/** @description Usage and cost info, populated when status is 'completed' */
 		TranscriptionUsage: {
@@ -816,7 +814,7 @@ export interface StreamWord {
 	start: number;
 	end: number;
 	confidence?: number;
-	speaker?: string | null;
+	speaker?: number | null;
 	punctuated_word?: string;
 }
 export interface SessionBeginMessage {
@@ -866,7 +864,7 @@ export type StreamEvents = {
 	};
 };
 export interface TranscribeStreamOptions {
-	model?: "audicle-o1" | "audicle-o1-realtime";
+	model?: "audicle-o1-realtime";
 	language?: string;
 	sample_rate?: number;
 	encoding?: "pcm_s16le";

package/dist/index.d.ts

@@ -34,16 +34,12 @@ export interface paths {
 		};
 		/**
 		 * Stream audio for live transcription
-		 * @description Initiates a WebSocket connection for live audio transcription.
-		 *
-		 *     The `model` query parameter determines the transcription backend:
-		 *     - `audicle-o1` — Whisper-based streaming (TranscriptionSession DO)
-		 *     - `audicle-o1-realtime` — streaming provider relay (StreamingSession DO)
+		 * @description Initiates a WebSocket connection for live audio transcription. Only streaming models (e.g. `audicle-o1-realtime`) are supported — prerecorded models must use `POST /v1/transcribe`.
 		 *
 		 *     **Protocol:** Send binary audio frames (PCM s16le) over the WebSocket. The server responds with JSON messages:
 		 *     - `{"type": "session.begin", "id": "...", "model": "..."}` — connection established
 		 *     - `{"type": "transcript", "is_final": true/false, "transcript": {"text": "...", "start": 0.0, "end": 1.5}}` — transcription results
-		 *     - `{"type": "vad", "event": "speech_start|speech_end", "timestamp_ms": N}` — voice activity detection (streaming models only)
+		 *     - `{"type": "vad", "event": "speech_start|speech_end", "timestamp_ms": N}` — voice activity detection
 		 *     - `{"type": "error", "code": "...", "message": "..."}` — error occurred
 		 *     - `{"type": "session.end", "reason": "...", "usage": {...}}` — session ended
 		 *
@@ -240,6 +236,7 @@ export interface components {
 			end: number;
 			text: string;
 			confidence?: number;
+			speaker?: number;
 			words?: components["schemas"]["TranscriptionWord"][];
 		};
 		TranscriptionWord: {
@@ -247,6 +244,7 @@ export interface components {
 			start: number;
 			end: number;
 			confidence?: number;
+			speaker?: number;
 		};
 		/** @description Usage and cost info, populated when status is 'completed' */
 		TranscriptionUsage: {
@@ -816,7 +814,7 @@ export interface StreamWord {
 	start: number;
 	end: number;
 	confidence?: number;
-	speaker?: string | null;
+	speaker?: number | null;
 	punctuated_word?: string;
 }
 export interface SessionBeginMessage {
@@ -866,7 +864,7 @@ export type StreamEvents = {
 	};
 };
 export interface TranscribeStreamOptions {
-	model?: "audicle-o1" | "audicle-o1-realtime";
+	model?: "audicle-o1-realtime";
 	language?: string;
 	sample_rate?: number;
 	encoding?: "pcm_s16le";

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@audicle/sdk",
-	"version": "0.1.1",
+	"version": "0.1.2",
 	"type": "module",
 	"description": "Official Node.js/TypeScript SDK for the Audicle transcription API",
 	"license": "MIT",