open-agents-ai 0.187.169 → 0.187.172

package/dist/index.js

@@ -258533,10 +258533,28 @@ var init_asr_listen = __esm({
           mkdirSync17(captureDir, { recursive: true });
         const audioFile = join51(captureDir, `listen-${Date.now()}.wav`);
         try {
-          execSync39(`arecord -D ${device} -f S16_LE -r 16000 -c 1 -d ${duration} -q ${audioFile}`, {
-            timeout: (duration + 5) * 1e3,
-            stdio: "pipe"
-          });
+          try {
+            execSync39(`which pw-record`, { stdio: "pipe", timeout: 2e3 });
+            const child = __require("child_process").spawnSync("pw-record", [
+              "--channels",
+              "1",
+              "--rate",
+              "16000",
+              "--format",
+              "s16",
+              audioFile
+            ], { timeout: (duration + 3) * 1e3, stdio: "pipe", killSignal: "SIGINT" });
+            execSync39(`pw-record --channels 1 --rate 16000 --format s16 ${audioFile} & PID=$!; sleep ${duration}; kill $PID 2>/dev/null; wait $PID 2>/dev/null`, {
+              timeout: (duration + 5) * 1e3,
+              stdio: "pipe",
+              shell: "/bin/bash"
+            });
+          } catch {
+            execSync39(`arecord -D ${device} -f S16_LE -r 16000 -c 1 -d ${duration} -q ${audioFile}`, {
+              timeout: (duration + 5) * 1e3,
+              stdio: "pipe"
+            });
+          }
         } catch (err) {
           return { success: false, output: "", error: `Recording failed: ${err instanceof Error ? err.message : String(err)}`, durationMs: performance.now() - start2 };
         }
@@ -258554,73 +258572,98 @@ var init_asr_listen = __esm({
         return this.doTranscribe(file, language, 0, start2);
       }
       doTranscribe(audioFile, language, recordDuration, start2) {
-        try {
-          const langArg = language !== "auto" ? `--language ${language}` : "";
-          const transcript = execSync39(`transcribe-cli ${langArg} "${audioFile}" 2>/dev/null`, { encoding: "utf8", timeout: 12e4 }).trim();
-          if (transcript && transcript.length > 0 && !transcript.includes("transcribe-cli")) {
-            return {
-              success: true,
-              output: `Transcription (${recordDuration > 0 ? recordDuration + "s recording, " : ""}Whisper):
+        const os8 = __require("node:os");
+        const fs4 = __require("node:fs");
+        const path5 = __require("node:path");
+        const homeDir = os8.homedir();
+        const tmpDir = os8.tmpdir();
+        const langArg = language !== "auto" ? `"${language}"` : "None";
+        const whisperScript = `
+import sys, json, os, warnings
+warnings.filterwarnings("ignore")
+os.environ["TF_CPP_MIN_LOG_LEVEL"] = "3"
 
-"${transcript}"
+audio_file = "${audioFile.replace(/\\/g, "/")}"
+lang = ${langArg}
 
-Audio file: ${audioFile}`,
-              durationMs: performance.now() - start2
-            };
-          }
-        } catch {
-        }
-        try {
-          const venvPy = join51(__require("os").homedir(), ".open-agents", "venv", "bin", "python3");
-          if (existsSync37(venvPy)) {
-            const script = `
-import sys
+# Try faster-whisper first (faster, lower memory)
 try:
     from faster_whisper import WhisperModel
-    model = WhisperModel("base", device="cpu", compute_type="int8")
-    segments, info = model.transcribe("${audioFile}", language="${language === "auto" ? "" : language}" or None)
-    text = " ".join(seg.text.strip() for seg in segments)
-    print(text)
-except Exception as e:
-    print(f"ERROR: {e}", file=sys.stderr)
-    sys.exit(1)
+    model = WhisperModel("base", device="auto", compute_type="int8")
+    segments, info = model.transcribe(audio_file, language=lang if lang else None)
+    text = " ".join(seg.text.strip() for seg in segments).strip()
+    if text:
+        print(json.dumps({"ok": True, "text": text, "backend": "faster-whisper"}))
+        sys.exit(0)
+except Exception:
+    pass
+
+# Try openai-whisper
+try:
+    import whisper
+    model = whisper.load_model("base")
+    result = model.transcribe(audio_file, language=lang if lang else None)
+    text = result.get("text", "").strip()
+    if text:
+        print(json.dumps({"ok": True, "text": text, "backend": "openai-whisper"}))
+        sys.exit(0)
+except Exception:
+    pass
+
+print(json.dumps({"ok": False, "error": "No whisper backend available"}))
 `;
-            const scriptFile = join51(__require("os").tmpdir(), `oa-whisper-${Date.now()}.py`);
-            __require("fs").writeFileSync(scriptFile, script);
-            const transcript = execSync39(`${venvPy} ${scriptFile}`, { encoding: "utf8", timeout: 12e4 }).trim();
-            if (transcript && !transcript.startsWith("ERROR")) {
-              return {
-                success: true,
-                output: `Transcription (faster-whisper):
-
-"${transcript}"
-
-Audio file: ${audioFile}`,
-                durationMs: performance.now() - start2
-              };
+        const pyPaths = [
+          path5.join(homeDir, ".open-agents", "venv", "bin", "python3"),
+          "python3",
+          "python"
+        ];
+        for (const pyPath of pyPaths) {
+          if (pyPath.includes("/") && !fs4.existsSync(pyPath))
+            continue;
+          const scriptFile = path5.join(tmpDir, `oa-asr-${Date.now()}.py`);
+          fs4.writeFileSync(scriptFile, whisperScript);
+          try {
+            const output = execSync39(`"${pyPath}" "${scriptFile}"`, {
+              encoding: "utf8",
+              timeout: 12e4,
+              env: { ...process.env, PYTHONUNBUFFERED: "1" }
+            }).trim();
+            try {
+              fs4.unlinkSync(scriptFile);
+            } catch {
             }
-          }
-        } catch {
-        }
-        try {
-          const transcript = execSync39(`whisper "${audioFile}" --language ${language} --model base --output_format txt 2>/dev/null`, { encoding: "utf8", timeout: 12e4 }).trim();
-          if (transcript) {
-            return {
-              success: true,
-              output: `Transcription (whisper):
+            const lines = output.split("\n");
+            for (let i2 = lines.length - 1; i2 >= 0; i2--) {
+              try {
+                const result = JSON.parse(lines[i2]);
+                if (result.ok && result.text) {
+                  return {
+                    success: true,
+                    output: `Transcription (${result.backend}, ${recordDuration > 0 ? recordDuration + "s recording" : "file"}):
 
-"${transcript}"
+"${result.text}"
 
 Audio file: ${audioFile}`,
-              durationMs: performance.now() - start2
-            };
+                    durationMs: performance.now() - start2
+                  };
+                }
+                if (!result.ok)
+                  continue;
+              } catch {
+                continue;
+              }
+            }
+          } catch {
+            try {
+              fs4.unlinkSync(scriptFile);
+            } catch {
+            }
           }
-        } catch {
         }
         return {
           success: false,
           output: `Audio recorded to: ${audioFile}`,
-          error: "Transcription failed \u2014 no Whisper backend available. Install: pip install faster-whisper",
+          error: "Transcription failed. Install whisper: pip install openai-whisper OR pip install faster-whisper",
           durationMs: performance.now() - start2
         };
       }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "open-agents-ai",
-  "version": "0.187.169",
+  "version": "0.187.172",
   "description": "AI coding agent powered by open-source models (Ollama/vLLM) — interactive TUI with agentic tool-calling loop",
   "type": "module",
   "main": "./dist/index.js",