pi-voice-input 0.3.0 → 0.3.2

package/README.md

@@ -18,9 +18,10 @@ Typing long prompts can slow you down. `pi-voice-input` lets you:
 - **One-key dictation**: `Ctrl+Shift+R` starts recording; press it again to stop and insert text.
 - **Editor-safe workflow**: transcription is pasted into the current editor only. It does not auto-submit.
 - **Chinese/English mixed input**: handles prompts that switch between Chinese, English, product names, and technical terms.
+- **Hotword table support**: can pass a VolcEngine boosting table ID to improve recognition of project terms.
 - **Works on Linux and macOS**: uses common system recording tools.
 - **Lowers sound while you speak**: automatically turns down system audio during recording, then restores it afterwards.
-- **No hidden rewriting**: inserts the raw ASR transcript, prefixed with a short note that it may contain voice-recognition errors.
+- **No hidden rewriting**: inserts the raw ASR transcript, prefixed with a short note that it may contain voice-recognition errors. If the editor already contains that note, later dictation inserts only the transcript.
 - **Simple setup commands**: configure from inside pi with `/voice init` and `/voice key`.
 
 Current speech provider: **VolcEngine Speech ASR**. A VolcEngine Speech API key is required.
@@ -98,19 +99,22 @@ Useful commands:
 
 ## Inserted text format
 
-The extension does not call a model to modify or translate your transcript. It inserts a short caveat telling the downstream model that the following text came from voice input and may contain recognition errors, then appends the raw ASR transcript unchanged.
+The extension does not call a model to modify or translate your transcript. It inserts a concise, location-neutral caveat saying the current conversation may include voice transcription errors, asking the model to correct them from context or ask the user if the meaning is unclear, then appends the raw ASR transcript unchanged. When you dictate multiple times in the same unsent editor draft, the caveat is kept to a single copy.
 
 User config keys are:
 
 ```json
 {
   "volcApiKey": "",
+  "boostingTableId": "",
   "duckSystemVolume": true,
   "duckSystemVolumeFactor": 0.5,
   "duckSystemVolumeFadeMs": 300
 }
 ```
 
+Set `boostingTableId` to a VolcEngine hotword/boosting table ID to send it as `boosting_table_id` on ASR requests. Leave it empty to disable hotword-table boosting. Boosting table name is not configured yet.
+
 ## System requirements
 
 Linux needs one recording tool:
@@ -135,7 +139,7 @@ On macOS, allow microphone access for your terminal or pi host app when prompted
 ## Troubleshooting
 
 - Run `/voice status` to see whether recording is active.
-- Run `/voice config` to confirm the API key is detected.
+- Run `/voice config` to confirm the API key and optional boosting table ID are detected.
 - Run `/voice key` again if the key was changed or expired.
 - On macOS, check microphone permission if recording immediately fails.
 - On Linux, make sure `pw-record` or `arecord` is installed and your microphone works in other apps.

package/ROADMAP.md

@@ -7,6 +7,7 @@ This roadmap lists user-visible work planned for pi Voice Input. It is intention
 - Linux voice input through `pw-record` or `arecord`
 - macOS voice input through the system `afrecord` command
 - VolcEngine WebSocket ASR
+- Optional VolcEngine hotword boosting table ID
 - Raw ASR transcript insertion with a voice-input caveat wrapper
 
 ## Planned

package/extensions/voice-input.ts

@@ -26,11 +26,7 @@ const VOLC_API_KEY_URL = "https://console.volcengine.com/speech/new/setting/apik
 // the local machine, so force those subprocesses to a known-local cwd.
 const LOCAL_AUDIO_PROCESS_CWD = homedir();
 const DEFAULT_SHORTCUT = Key.ctrlShift("r");
-const VOICE_TRANSCRIPT_NOTICE = [
-  "以下内容来自用户通过语音输入的原始转写，可能包含语音识别错误、错别字、同音词、断句或标点不准确，以及中英文、术语、代码名、文件名误识别等问题。",
-  "请明确注意：这段转写不一定完全准确。请结合上下文理解用户意图；必要时可以先澄清、翻译或推断，再继续处理。",
-  "原始语音转写如下：",
-].join("\n");
+const VOICE_TRANSCRIPT_NOTICE = "当前会话中的内容包含语音转写，可能存在识别错误；请结合上下文纠正，若不确定或明显偏离主题无法理解，请先询问用户。";
 
 const MSG_TYPE_CLIENT_FULL_REQUEST = 0b0001;
 const MSG_TYPE_CLIENT_AUDIO_ONLY_REQUEST = 0b0010;
@@ -46,6 +42,7 @@ type JsonObject = Record<string, unknown>;
 
 type VoiceInputConfigFile = {
   volcApiKey: string;
+  boostingTableId: string;
   duckSystemVolume: boolean;
   duckSystemVolumeFactor: number;
   duckSystemVolumeFadeMs: number;
@@ -59,6 +56,7 @@ type VoiceConfig = {
   language: string;
   uid: string;
   prompt: string;
+  boostingTableId: string;
   segmentMs: number;
   requestTimeoutMs: number;
   finalizeDelayMs: number;
@@ -118,6 +116,7 @@ function ensureDir(dir: string) {
 function defaultConfigFile(): VoiceInputConfigFile {
   return {
     volcApiKey: "",
+    boostingTableId: "",
     duckSystemVolume: true,
     duckSystemVolumeFactor: 0.5,
     duckSystemVolumeFadeMs: 300,
@@ -152,6 +151,11 @@ function normalizeConfigFile(input: unknown): VoiceInputConfigFile {
   const root = isObject(input) ? input : {};
   return {
     volcApiKey: stringField(root, "volcApiKey", defaults.volcApiKey).trim(),
+    boostingTableId: stringField(
+      root,
+      "boostingTableId",
+      stringField(root, "boosting_table_id", defaults.boostingTableId),
+    ).trim(),
     duckSystemVolume: booleanField(root, "duckSystemVolume", defaults.duckSystemVolume),
     duckSystemVolumeFactor: clamp(numberField(root, "duckSystemVolumeFactor", defaults.duckSystemVolumeFactor), 0, 1),
     duckSystemVolumeFadeMs: Math.round(clamp(numberField(root, "duckSystemVolumeFadeMs", defaults.duckSystemVolumeFadeMs), 0, 3000)),
@@ -185,6 +189,7 @@ function getConfig(): VoiceConfig {
     language: "",
     uid: "pi-voice-input",
     prompt: "",
+    boostingTableId: fileConfig.boostingTableId,
     segmentMs: 5000,
     requestTimeoutMs: 90000,
     finalizeDelayMs: 100,
@@ -794,6 +799,7 @@ async function transcribePcm(pcm: Buffer, durationMs: number, config: VoiceConfi
       "X-Api-Resource-Id": config.resourceId,
       "X-Api-Connect-Id": connectId,
       "X-Api-Request-Id": connectId,
+      "X-Api-Sequence": "-1",
     },
     handshakeTimeout: 15_000,
   });
@@ -871,6 +877,10 @@ async function transcribePcm(pcm: Buffer, durationMs: number, config: VoiceConfi
     audioPayload.language = config.language;
   }
 
+  const corpusPayload: Record<string, unknown> = {};
+  if (config.boostingTableId) corpusPayload.boosting_table_id = config.boostingTableId;
+  if (config.prompt) corpusPayload.context = config.prompt;
+
   const requestPayload: Record<string, unknown> = {
     user: { uid: config.uid || "pi-voice-input" },
     audio: audioPayload,
@@ -881,7 +891,7 @@ async function transcribePcm(pcm: Buffer, durationMs: number, config: VoiceConfi
       enable_ddc: config.enableDdc,
       show_utterances: config.showUtterances,
       result_type: "full",
-      ...(config.prompt ? { corpus: { context: config.prompt } } : {}),
+      ...(Object.keys(corpusPayload).length ? { corpus: corpusPayload } : {}),
     },
   };
 
@@ -928,9 +938,14 @@ async function transcribePcm(pcm: Buffer, durationMs: number, config: VoiceConfi
   };
 }
 
-function wrapVoiceTranscript(rawText: string): string {
+function editorHasVoiceTranscriptNotice(editorText: string): boolean {
+  return editorText.replace(/\r\n/g, "\n").includes(VOICE_TRANSCRIPT_NOTICE);
+}
+
+function wrapVoiceTranscript(rawText: string, currentEditorText = ""): string {
   const trimmed = rawText.trim();
   if (!trimmed) return "";
+  if (editorHasVoiceTranscriptNotice(currentEditorText)) return trimmed;
   return `${VOICE_TRANSCRIPT_NOTICE}
 
 ${trimmed}`;
@@ -1084,7 +1099,7 @@ async function stopRecording(ctx: ExtensionContext, transcribe = true) {
     return;
   }
 
-  const finalText = wrapVoiceTranscript(result.text);
+  const finalText = wrapVoiceTranscript(result.text, ctx.ui.getEditorText());
 
   ctx.ui.setStatus("voice-input", undefined);
   insertIntoEditor(ctx, finalText);
@@ -1107,6 +1122,7 @@ function setupHelp(config = getConfig()): string {
     "- Current provider: VolcEngine WebSocket ASR",
     `- Config file: ${config.configPath}`,
     `- API key: ${config.apiKey ? "set" : "missing"}`,
+    `- Hotword boosting table ID: ${config.boostingTableId ? "set" : "not set"}`,
     "- To create/update the JSON config file, run: /voice init",
     "- To save/update the key, run: /voice key",
     "- Output: raw ASR transcript wrapped with a short voice-input caveat",
@@ -1146,12 +1162,13 @@ function configSummary(config: VoiceConfig): string {
     "Voice input config:",
     `- config file: ${config.configPath}${existsSync(config.configPath) ? "" : " (missing; run /voice init to create it)"}`,
     `- volcApiKey: ${config.apiKey ? "set" : "missing"} (update with /voice key)`,
+    `- boostingTableId: ${config.boostingTableId ? "set" : "not set"}`,
     "- outputMode: raw transcript with voice-input caveat wrapper",
     `- duckSystemVolume: ${config.duckSystemVolume ? "enabled" : "disabled"}`,
     `- duckSystemVolumeFactor: ${config.duckSystemVolumeFactor}`,
     `- duckSystemVolumeFadeMs: ${config.duckSystemVolumeFadeMs}`,
     `- current recording device: ${currentDevice}`,
-    "Config keys: volcApiKey, duckSystemVolume, duckSystemVolumeFactor, duckSystemVolumeFadeMs.",
+    "Config keys: volcApiKey, boostingTableId, duckSystemVolume, duckSystemVolumeFactor, duckSystemVolumeFadeMs.",
     `VolcEngine API key URL: ${VOLC_API_KEY_URL}`,
   ].join("\n");
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pi-voice-input",
-  "version": "0.3.0",
+  "version": "0.3.2",
   "description": "Press Ctrl+Shift+R to dictate prompts into Pi using VolcEngine ASR",
   "type": "module",
   "keywords": [