@pipedog/openai 0.4.0 → 0.5.0

package/README.md

@@ -0,0 +1,53 @@
+# @pipedog/openai
+
+OpenAI services for pipedog: the chat completions LLM (`OpenAILLMService`), the Responses API
+(`OpenAIResponsesLLMService`), Realtime (`OpenAIRealtimeLLMService`), Live, speech-to-text
+(`OpenAISTTService`, and `OpenAIRealtimeSTTService` with server-side turn detection),
+text-to-speech (`OpenAITTSService`) and image generation (`OpenAIImageGenService`). It uses the
+official `openai` JS SDK (installed as a dependency), and `baseUrl` points it at OpenAI-compatible
+servers.
+
+A port of Pipecat's `pipecat.services.openai` and the OpenAI adapters in
+`pipecat.adapters.services`. Subpaths mirror upstream: `@pipedog/openai/llm`, `/stt`, `/tts`,
+`/realtime/llm`, `/live/llm`, `/responses/llm`, `/image`.
+
+## Install
+
+```sh
+bun add @pipedog/core @pipedog/asyncio @pipedog/openai
+```
+
+It depends on [`@pipedog/whisper`](https://www.npmjs.com/package/@pipedog/whisper), which is
+installed with it.
+
+## Usage
+
+```ts
+import { OpenAILLMSettings } from "@pipedog/openai/base_llm";
+import { OpenAILLMService } from "@pipedog/openai/llm";
+import { OpenAISTTService } from "@pipedog/openai/stt";
+import { OpenAITTSService } from "@pipedog/openai/tts";
+
+const apiKey = process.env.OPENAI_API_KEY ?? null;
+
+const stt = new OpenAISTTService({ apiKey });
+const llm = new OpenAILLMService({
+  apiKey,
+  settings: new OpenAILLMSettings({ systemInstruction: "You are a helpful voice assistant." }),
+});
+const tts = new OpenAITTSService({ apiKey, sampleRate: 24000 });
+// new Pipeline([transport.input(), stt, userAggregator, llm, tts, transport.output(), ...])
+```
+
+## Runtimes
+
+Bun, Node.js 22+, Deno and Cloudflare Workers.
+
+## Links
+
+- Docs: [voice bot guide](https://polvallverdu.github.io/pipedog/docs/voice-bot/),
+  [turn detection](https://polvallverdu.github.io/pipedog/docs/turn-detection/) (realtime STT turns),
+  [Workers](https://polvallverdu.github.io/pipedog/docs/workers/)
+- [pipedog README](https://github.com/polvallverdu/pipedog#readme)
+- License: BSD 2-Clause, a port of Pipecat (Copyright (c) 2024–2026 Daily). See
+  [LICENSE](https://github.com/polvallverdu/pipedog/blob/main/LICENSE).

package/dist/live/llm.d.ts

@@ -3,9 +3,10 @@
  *
  * Porting notes:
  * - The WebSocket is opened with `websocketConnect` from core, which sends
- *   the `Authorization` header on Bun, Node and Deno. Runtimes whose
- *   `WebSocket` can't send headers (Cloudflare Workers) pass a
- *   `webSocketFactory` (an addition to the upstream options).
+ *   the `Authorization` header on Bun, Node, Deno and Cloudflare Workers
+ *   (through `fetch` with an `Upgrade: websocket` header). Other runtimes
+ *   whose `WebSocket` can't send headers pass a `webSocketFactory` (an
+ *   addition to the upstream options).
  * - `parse_server_event` raises only `ValueError`, which the receive loop
  *   skips; here any (non-cancellation) error from `parseServerEvent` is
  *   treated as that.
@@ -204,7 +205,8 @@ export interface OpenAILiveLLMServiceOptions extends LLMServiceOptions {
     delegation?: ResponsesDelegation | ClientDelegation | null;
     /**
      * Creates the underlying `WebSocket` (see `WebsocketConnectOptions`), for
-     * runtimes whose `WebSocket` can't send the `Authorization` header.
+     * runtimes other than Bun, Node, Deno and Cloudflare Workers, whose
+     * `WebSocket` can't send the `Authorization` header.
      */
     webSocketFactory?: WebsocketConnectOptions["webSocketFactory"];
 }

package/dist/live/llm.js

@@ -8,9 +8,10 @@
  *
  * Porting notes:
  * - The WebSocket is opened with `websocketConnect` from core, which sends
- *   the `Authorization` header on Bun, Node and Deno. Runtimes whose
- *   `WebSocket` can't send headers (Cloudflare Workers) pass a
- *   `webSocketFactory` (an addition to the upstream options).
+ *   the `Authorization` header on Bun, Node, Deno and Cloudflare Workers
+ *   (through `fetch` with an `Upgrade: websocket` header). Other runtimes
+ *   whose `WebSocket` can't send headers pass a `webSocketFactory` (an
+ *   addition to the upstream options).
  * - `parse_server_event` raises only `ValueError`, which the receive loop
  *   skips; here any (non-cancellation) error from `parseServerEvent` is
  *   treated as that.

package/dist/stt.d.ts

@@ -13,9 +13,9 @@
  *   `BaseWhisperSTTService` (`@pipedog/whisper`).
  * - `OpenAIRealtimeSTTService` uses the global `WebSocket` through
  *   `WebsocketService._websocketConnect()`. The `Authorization` header is sent
- *   on the opening handshake, which the Bun, Node and Deno `WebSocket`
- *   support; on Cloudflare Workers override `_connectWebsocket()` or the
- *   connect options to open the socket through `fetch`.
+ *   on the opening handshake on Bun, Node, Deno and Cloudflare Workers (where
+ *   core opens the socket through `fetch` with an `Upgrade: websocket`
+ *   header); browsers can't send it.
  * - Deprecated APIs are not ported: the `model`, `language`, `prompt`,
  *   `temperature` (`OpenAISTTService`) and `model`, `language`, `prompt`,
  *   `noiseReduction` (`OpenAIRealtimeSTTService`) constructor options.

package/dist/stt.js

@@ -18,9 +18,9 @@
  *   `BaseWhisperSTTService` (`@pipedog/whisper`).
  * - `OpenAIRealtimeSTTService` uses the global `WebSocket` through
  *   `WebsocketService._websocketConnect()`. The `Authorization` header is sent
- *   on the opening handshake, which the Bun, Node and Deno `WebSocket`
- *   support; on Cloudflare Workers override `_connectWebsocket()` or the
- *   connect options to open the socket through `fetch`.
+ *   on the opening handshake on Bun, Node, Deno and Cloudflare Workers (where
+ *   core opens the socket through `fetch` with an `Upgrade: websocket`
+ *   header); browsers can't send it.
  * - Deprecated APIs are not ported: the `model`, `language`, `prompt`,
  *   `temperature` (`OpenAISTTService`) and `model`, `language`, `prompt`,
  *   `noiseReduction` (`OpenAIRealtimeSTTService`) constructor options.

package/package.json

@@ -1,7 +1,19 @@
 {
   "name": "@pipedog/openai",
-  "version": "0.4.0",
+  "version": "0.5.0",
   "description": "OpenAI services (LLM, Responses, Realtime, Live, STT, TTS, image) for pipedog.",
+  "keywords": [
+    "pipecat",
+    "pipedog",
+    "voice-ai",
+    "conversational-ai",
+    "realtime",
+    "openai",
+    "llm",
+    "stt",
+    "tts",
+    "realtime-api"
+  ],
   "license": "BSD-2-Clause",
   "repository": {
     "type": "git",
@@ -33,9 +45,9 @@
     "test": "bun test"
   },
   "dependencies": {
-    "@pipedog/asyncio": "0.4.0",
-    "@pipedog/core": "0.4.0",
-    "@pipedog/whisper": "0.4.0",
+    "@pipedog/asyncio": "0.5.0",
+    "@pipedog/core": "0.5.0",
+    "@pipedog/whisper": "0.5.0",
     "fast-png": "^8.0.0",
     "openai": "^6.49.0"
   },