Add voice audio, capture and transcript, browser monitor, and DM UI

poc/src/stt.ts

@@ -0,0 +1,135 @@
+/*
+ * Veskslop daemon — speech to text.
+ *
+ * Providers, in order: VESKSLOP_STT_CMD, a local Vosk model
+ * (VESKSLOP_VOSK_MODEL or a `vosk-model*` folder under poc/models), and an
+ * HTTP API. The local command receives the WAV path ({wav}) and prints text.
+ */
+
+import { spawn } from "node:child_process";
+import { readdirSync, readFileSync } from "node:fs";
+import { join } from "node:path";
+import { fileURLToPath } from "node:url";
+
+export class SttError extends Error {}
+
+export interface SttStatus {
+    local: boolean;
+    api: boolean;
+    engine: string | null;
+    hint?: string;
+}
+
+export class Stt {
+    private voskModel?: string;
+
+    get localCommand(): string | undefined {
+        return process.env.VESKSLOP_STT_CMD;
+    }
+
+    get apiUrl(): string | undefined {
+        return process.env.VESKSLOP_STT_API_URL;
+    }
+
+    async detect(): Promise<void> {
+        this.voskModel = this.findVoskModel();
+    }
+
+    status(): SttStatus {
+        const local = !!(this.localCommand || this.voskModel);
+
+        return {
+            local,
+            api: !!this.apiUrl,
+            engine: this.localCommand ? "command" : this.voskModel ? "vosk" : null,
+            hint:
+                local || this.apiUrl
+                    ? undefined
+                    : "Install vosk (pip install vosk) and set VESKSLOP_VOSK_MODEL, or set VESKSLOP_STT_CMD."
+        };
+    }
+
+    /** Returns undefined when no provider is configured. */
+    async transcribe(wavPath: string): Promise<string | undefined> {
+        if (this.localCommand) return this.viaLocalCommand(wavPath, this.localCommand);
+
+        if (this.voskModel) {
+            const script = fileURLToPath(new URL("../scripts/stt-vosk.py", import.meta.url));
+            return this.viaLocalCommand(wavPath, `python3 ${script} {wav}`, {
+                VESKSLOP_VOSK_MODEL: this.voskModel
+            });
+        }
+
+        if (this.apiUrl) return this.viaApi(wavPath);
+        return undefined;
+    }
+
+    private findVoskModel(): string | undefined {
+        const explicit = process.env.VESKSLOP_VOSK_MODEL;
+        if (explicit) return explicit;
+
+        const base = process.env.VESKSLOP_VOSK_DIR ?? fileURLToPath(new URL("../models", import.meta.url));
+
+        try {
+            const model = readdirSync(base, { withFileTypes: true }).find(
+                entry => entry.isDirectory() && entry.name.startsWith("vosk-model")
+            );
+            return model ? join(base, model.name) : undefined;
+        } catch {
+            return undefined;
+        }
+    }
+
+    private async viaLocalCommand(
+        wavPath: string,
+        command: string,
+        extraEnv: Record<string, string> = {}
+    ): Promise<string> {
+        if (!command.includes("{wav}")) throw new SttError("the STT command must contain {wav}");
+
+        const parts = command
+            .split(/\s+/)
+            .filter(Boolean)
+            .map(part => part.replace("{wav}", wavPath));
+
+        return new Promise<string>((resolve, reject) => {
+            const child = spawn(parts[0], parts.slice(1), {
+                env: { ...process.env, ...extraEnv },
+                stdio: ["ignore", "pipe", "pipe"]
+            });
+            let stdout = "";
+            let stderr = "";
+
+            child.stdout?.on("data", chunk => (stdout += chunk));
+            child.stderr?.on("data", chunk => (stderr += chunk));
+            child.on("error", error => reject(new SttError(String(error))));
+            child.on("close", code => {
+                if (code === 0) resolve(stdout.trim());
+                else reject(new SttError(stderr.trim() || `the STT command exited with ${code}`));
+            });
+        });
+    }
+
+    private async viaApi(wavPath: string): Promise<string> {
+        const response = await fetch(this.apiUrl!, {
+            method: "POST",
+            headers: {
+                "Content-Type": "audio/wav",
+                ...(process.env.VESKSLOP_STT_API_KEY
+                    ? { Authorization: `Bearer ${process.env.VESKSLOP_STT_API_KEY}` }
+                    : {})
+            },
+            body: readFileSync(wavPath)
+        });
+
+        if (!response.ok) throw new SttError(`the STT API returned ${response.status}`);
+
+        const type = response.headers.get("content-type") ?? "";
+        if (type.includes("json")) {
+            const data: any = await response.json();
+            return String(data.text ?? data.transcription ?? "").trim();
+        }
+
+        return (await response.text()).trim();
+    }
+}