@hasna/recordings 0.1.32 → 0.1.33

package/dist/cli/index.js

@@ -5003,7 +5003,7 @@ var init_pg_migrations = __esm(() => {
     raw_text TEXT NOT NULL,
     processed_text TEXT,
     processing_mode TEXT NOT NULL DEFAULT 'raw' CHECK(processing_mode IN ('raw', 'enhanced')),
-    model_used TEXT NOT NULL DEFAULT 'gpt-4o-mini-transcribe',
+    model_used TEXT NOT NULL DEFAULT 'gpt-4o-transcribe',
     enhancement_model TEXT,
     duration_ms INTEGER DEFAULT 0,
     language TEXT,
@@ -5098,7 +5098,7 @@ var init_pg_migrate = __esm(() => {
 var require_package = __commonJS((exports, module) => {
   module.exports = {
     name: "@hasna/recordings",
-    version: "0.1.32",
+    version: "0.1.33",
     type: "module",
     description: "Speech-to-text recording tool with MCP and CLI \u2014 records, transcribes, and optionally enhances text using AI",
     repository: {
@@ -5865,7 +5865,7 @@ import { homedir as homedir2 } from "os";
 var DEFAULT_CONFIG = {
   openai_api_key: "",
   enhancement_api_key: "",
-  transcription_model: "gpt-4o-mini-transcribe",
+  transcription_model: "gpt-4o-transcribe",
   enhancement_model: "gpt-4o",
   language: "en",
   audio_format: "wav",
@@ -6103,7 +6103,7 @@ var MIGRATIONS = [
     raw_text TEXT NOT NULL,
     processed_text TEXT,
     processing_mode TEXT NOT NULL DEFAULT 'raw' CHECK(processing_mode IN ('raw', 'enhanced')),
-    model_used TEXT NOT NULL DEFAULT 'gpt-4o-mini-transcribe',
+    model_used TEXT NOT NULL DEFAULT 'gpt-4o-transcribe',
     enhancement_model TEXT,
     duration_ms INTEGER DEFAULT 0,
     language TEXT,
@@ -6233,7 +6233,7 @@ function parseRow(row) {
     raw_text: row["raw_text"],
     processed_text: row["processed_text"] || null,
     processing_mode: row["processing_mode"] || "raw",
-    model_used: row["model_used"] || "gpt-4o-mini-transcribe",
+    model_used: row["model_used"] || "gpt-4o-transcribe",
     enhancement_model: row["enhancement_model"] || null,
     duration_ms: row["duration_ms"] || 0,
     language: row["language"] || null,
@@ -6255,7 +6255,7 @@ function createRecording(input, db) {
   const tagsJson = JSON.stringify(input.tags || []);
   const metadataJson = JSON.stringify(input.metadata || {});
   d.query(`INSERT INTO recordings (id, audio_path, raw_text, processed_text, processing_mode, model_used, enhancement_model, duration_ms, language, tags, agent_id, project_id, session_id, goal, role, task_list_id, machine_id, metadata)
-     VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`).run(id, input.audio_path || null, input.raw_text, input.processed_text || null, input.processing_mode || "raw", input.model_used || "gpt-4o-mini-transcribe", input.enhancement_model || null, input.duration_ms || 0, input.language || null, tagsJson, input.agent_id || null, input.project_id || null, input.session_id || null, input.goal || null, input.role || null, input.task_list_id || null, input.machine_id || null, metadataJson);
+     VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`).run(id, input.audio_path || null, input.raw_text, input.processed_text || null, input.processing_mode || "raw", input.model_used || "gpt-4o-transcribe", input.enhancement_model || null, input.duration_ms || 0, input.language || null, tagsJson, input.agent_id || null, input.project_id || null, input.session_id || null, input.goal || null, input.role || null, input.task_list_id || null, input.machine_id || null, metadataJson);
   if (input.tags && input.tags.length > 0) {
     const insertTag = d.query("INSERT OR IGNORE INTO recording_tags (recording_id, tag) VALUES (?, ?)");
     for (const tag of input.tags) {
@@ -6762,7 +6762,7 @@ async function processText(rawText, config, systemPrompt) {
 }
 
 // src/version.ts
-var VERSION = "0.1.32";
+var VERSION = "0.1.33";
 
 // src/cli/storage.ts
 import chalk from "chalk";
@@ -7415,7 +7415,7 @@ program.command("init").description("Initialize .recordings/ in current director
   mkdirSync3(audioDir, { recursive: true });
   if (!existsSync6(configFile)) {
     const defaultConf = {
-      transcription_model: "gpt-4o-mini-transcribe",
+      transcription_model: "gpt-4o-transcribe",
       enhancement_model: "gpt-4o",
       language: "en",
       auto_enhance: true

package/dist/index.js

@@ -4972,7 +4972,7 @@ import { homedir } from "os";
 var DEFAULT_CONFIG = {
   openai_api_key: "",
   enhancement_api_key: "",
-  transcription_model: "gpt-4o-mini-transcribe",
+  transcription_model: "gpt-4o-transcribe",
   enhancement_model: "gpt-4o",
   language: "en",
   audio_format: "wav",
@@ -5177,7 +5177,7 @@ var MIGRATIONS = [
     raw_text TEXT NOT NULL,
     processed_text TEXT,
     processing_mode TEXT NOT NULL DEFAULT 'raw' CHECK(processing_mode IN ('raw', 'enhanced')),
-    model_used TEXT NOT NULL DEFAULT 'gpt-4o-mini-transcribe',
+    model_used TEXT NOT NULL DEFAULT 'gpt-4o-transcribe',
     enhancement_model TEXT,
     duration_ms INTEGER DEFAULT 0,
     language TEXT,
@@ -5460,7 +5460,7 @@ var PG_MIGRATIONS = [
     raw_text TEXT NOT NULL,
     processed_text TEXT,
     processing_mode TEXT NOT NULL DEFAULT 'raw' CHECK(processing_mode IN ('raw', 'enhanced')),
-    model_used TEXT NOT NULL DEFAULT 'gpt-4o-mini-transcribe',
+    model_used TEXT NOT NULL DEFAULT 'gpt-4o-transcribe',
     enhancement_model TEXT,
     duration_ms INTEGER DEFAULT 0,
     language TEXT,
@@ -5701,7 +5701,7 @@ function parseRow(row) {
     raw_text: row["raw_text"],
     processed_text: row["processed_text"] || null,
     processing_mode: row["processing_mode"] || "raw",
-    model_used: row["model_used"] || "gpt-4o-mini-transcribe",
+    model_used: row["model_used"] || "gpt-4o-transcribe",
     enhancement_model: row["enhancement_model"] || null,
     duration_ms: row["duration_ms"] || 0,
     language: row["language"] || null,
@@ -5723,7 +5723,7 @@ function createRecording(input, db) {
   const tagsJson = JSON.stringify(input.tags || []);
   const metadataJson = JSON.stringify(input.metadata || {});
   d.query(`INSERT INTO recordings (id, audio_path, raw_text, processed_text, processing_mode, model_used, enhancement_model, duration_ms, language, tags, agent_id, project_id, session_id, goal, role, task_list_id, machine_id, metadata)
-     VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`).run(id, input.audio_path || null, input.raw_text, input.processed_text || null, input.processing_mode || "raw", input.model_used || "gpt-4o-mini-transcribe", input.enhancement_model || null, input.duration_ms || 0, input.language || null, tagsJson, input.agent_id || null, input.project_id || null, input.session_id || null, input.goal || null, input.role || null, input.task_list_id || null, input.machine_id || null, metadataJson);
+     VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`).run(id, input.audio_path || null, input.raw_text, input.processed_text || null, input.processing_mode || "raw", input.model_used || "gpt-4o-transcribe", input.enhancement_model || null, input.duration_ms || 0, input.language || null, tagsJson, input.agent_id || null, input.project_id || null, input.session_id || null, input.goal || null, input.role || null, input.task_list_id || null, input.machine_id || null, metadataJson);
   if (input.tags && input.tags.length > 0) {
     const insertTag = d.query("INSERT OR IGNORE INTO recording_tags (recording_id, tag) VALUES (?, ?)");
     for (const tag of input.tags) {

package/dist/mcp/index.js

@@ -4920,7 +4920,7 @@ var require_lib2 = __commonJS((exports, module) => {
 var require_package = __commonJS((exports, module) => {
   module.exports = {
     name: "@hasna/recordings",
-    version: "0.1.32",
+    version: "0.1.33",
     type: "module",
     description: "Speech-to-text recording tool with MCP and CLI \u2014 records, transcribes, and optionally enhances text using AI",
     repository: {
@@ -9021,7 +9021,7 @@ import { homedir } from "os";
 var DEFAULT_CONFIG = {
   openai_api_key: "",
   enhancement_api_key: "",
-  transcription_model: "gpt-4o-mini-transcribe",
+  transcription_model: "gpt-4o-transcribe",
   enhancement_model: "gpt-4o",
   language: "en",
   audio_format: "wav",
@@ -9259,7 +9259,7 @@ var MIGRATIONS = [
     raw_text TEXT NOT NULL,
     processed_text TEXT,
     processing_mode TEXT NOT NULL DEFAULT 'raw' CHECK(processing_mode IN ('raw', 'enhanced')),
-    model_used TEXT NOT NULL DEFAULT 'gpt-4o-mini-transcribe',
+    model_used TEXT NOT NULL DEFAULT 'gpt-4o-transcribe',
     enhancement_model TEXT,
     duration_ms INTEGER DEFAULT 0,
     language TEXT,
@@ -9389,7 +9389,7 @@ function parseRow(row) {
     raw_text: row["raw_text"],
     processed_text: row["processed_text"] || null,
     processing_mode: row["processing_mode"] || "raw",
-    model_used: row["model_used"] || "gpt-4o-mini-transcribe",
+    model_used: row["model_used"] || "gpt-4o-transcribe",
     enhancement_model: row["enhancement_model"] || null,
     duration_ms: row["duration_ms"] || 0,
     language: row["language"] || null,
@@ -9411,7 +9411,7 @@ function createRecording(input, db) {
   const tagsJson = JSON.stringify(input.tags || []);
   const metadataJson = JSON.stringify(input.metadata || {});
   d.query(`INSERT INTO recordings (id, audio_path, raw_text, processed_text, processing_mode, model_used, enhancement_model, duration_ms, language, tags, agent_id, project_id, session_id, goal, role, task_list_id, machine_id, metadata)
-     VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`).run(id, input.audio_path || null, input.raw_text, input.processed_text || null, input.processing_mode || "raw", input.model_used || "gpt-4o-mini-transcribe", input.enhancement_model || null, input.duration_ms || 0, input.language || null, tagsJson, input.agent_id || null, input.project_id || null, input.session_id || null, input.goal || null, input.role || null, input.task_list_id || null, input.machine_id || null, metadataJson);
+     VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`).run(id, input.audio_path || null, input.raw_text, input.processed_text || null, input.processing_mode || "raw", input.model_used || "gpt-4o-transcribe", input.enhancement_model || null, input.duration_ms || 0, input.language || null, tagsJson, input.agent_id || null, input.project_id || null, input.session_id || null, input.goal || null, input.role || null, input.task_list_id || null, input.machine_id || null, metadataJson);
   if (input.tags && input.tags.length > 0) {
     const insertTag = d.query("INSERT OR IGNORE INTO recording_tags (recording_id, tag) VALUES (?, ?)");
     for (const tag of input.tags) {
@@ -9798,7 +9798,7 @@ async function processText(rawText, config, systemPrompt) {
 }
 
 // src/version.ts
-var VERSION = "0.1.32";
+var VERSION = "0.1.33";
 
 // src/db/storage-config.ts
 import { existsSync as existsSync2, readFileSync as readFileSync2 } from "fs";
@@ -9957,7 +9957,7 @@ var PG_MIGRATIONS = [
     raw_text TEXT NOT NULL,
     processed_text TEXT,
     processing_mode TEXT NOT NULL DEFAULT 'raw' CHECK(processing_mode IN ('raw', 'enhanced')),
-    model_used TEXT NOT NULL DEFAULT 'gpt-4o-mini-transcribe',
+    model_used TEXT NOT NULL DEFAULT 'gpt-4o-transcribe',
     enhancement_model TEXT,
     duration_ms INTEGER DEFAULT 0,
     language TEXT,

package/dist/storage.js

@@ -5024,7 +5024,7 @@ import { homedir as homedir2 } from "os";
 var DEFAULT_CONFIG = {
   openai_api_key: "",
   enhancement_api_key: "",
-  transcription_model: "gpt-4o-mini-transcribe",
+  transcription_model: "gpt-4o-transcribe",
   enhancement_model: "gpt-4o",
   language: "en",
   audio_format: "wav",
@@ -5229,7 +5229,7 @@ var MIGRATIONS = [
     raw_text TEXT NOT NULL,
     processed_text TEXT,
     processing_mode TEXT NOT NULL DEFAULT 'raw' CHECK(processing_mode IN ('raw', 'enhanced')),
-    model_used TEXT NOT NULL DEFAULT 'gpt-4o-mini-transcribe',
+    model_used TEXT NOT NULL DEFAULT 'gpt-4o-transcribe',
     enhancement_model TEXT,
     duration_ms INTEGER DEFAULT 0,
     language TEXT,
@@ -5434,7 +5434,7 @@ var PG_MIGRATIONS = [
     raw_text TEXT NOT NULL,
     processed_text TEXT,
     processing_mode TEXT NOT NULL DEFAULT 'raw' CHECK(processing_mode IN ('raw', 'enhanced')),
-    model_used TEXT NOT NULL DEFAULT 'gpt-4o-mini-transcribe',
+    model_used TEXT NOT NULL DEFAULT 'gpt-4o-transcribe',
     enhancement_model TEXT,
     duration_ms INTEGER DEFAULT 0,
     language TEXT,

package/dist/version.d.ts

@@ -1,2 +1,2 @@
-export declare const VERSION = "0.1.32";
+export declare const VERSION = "0.1.33";
 //# sourceMappingURL=version.d.ts.map

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@hasna/recordings",
-  "version": "0.1.32",
+  "version": "0.1.33",
   "type": "module",
   "description": "Speech-to-text recording tool with MCP and CLI — records, transcribes, and optionally enhances text using AI",
   "repository": {

package/src/native/Recordings/RecordingsLib/RecordingEngine.swift

@@ -520,7 +520,6 @@ public final class RecordingEngine: ObservableObject {
 
     public func stopAndTranscribe() {
         guard isRecording else { return }
-        let stopStartedAt = Date()
         log("stopAndTranscribe")
 
         recordingTimer?.invalidate()
@@ -541,7 +540,6 @@ public final class RecordingEngine: ObservableObject {
         let audioPath = activeAudioPath
         let pcmStreamPipe = pcmStreamPipe
         let client = realtimeClient
-        let transcriptionLanguage = transcriptionLanguage
         resetRecordingIntent()
         self.pcmStreamPipe = nil
 
@@ -558,37 +556,11 @@ public final class RecordingEngine: ObservableObject {
             self.streamingTask = nil
 
             let realtimeText = Self.normalizedRealtimeTranscript(streamingResult)
-            let fastPathText = Self.realtimeFastPathTranscript(
-                realtimeText: streamingResult,
-                pcmByteCount: self.recordedPCM.count,
-                language: transcriptionLanguage
-            )
 
             self.liveTranscriptionText = ""
 
-            if let fastPathText {
-                let releaseToTextMS = Int(Date().timeIntervalSince(stopStartedAt) * 1_000)
-                let repaired = Self.wasRealtimeTranscriptRepaired(rawText: streamingResult, cleanedText: fastPathText)
-                self.log("using realtime fast path chars=\(fastPathText.count) repaired=\(repaired) releaseToTextMs=\(releaseToTextMS)")
-                self.isTranscribing = false
-                self.finishWithText(
-                    fastPathText,
-                    curMode: curMode,
-                    targetAppBundleIdentifier: targetAppBundleIdentifier,
-                    targetAppPid: targetAppPid,
-                    activeProjectId: activeProjectId,
-                    activeProjectName: activeProjectName
-                )
-                if let audioPath, !self.recordedPCM.isEmpty {
-                    Self.saveCapturedWAVInBackground(pcmData: self.recordedPCM, audioPath: audioPath, homePath: self.home)
-                }
-                self.activeAudioPath = nil
-                self.recordedPCM.removeAll(keepingCapacity: true)
-                return
-            }
-
             if let audioPath, self.writeCapturedWAV(to: audioPath) {
-                self.log("transcribing captured full audio audioPath=\(audioPath) realtimePreviewChars=\(realtimeText?.count ?? 0)")
+                self.log("transcribing captured full audio with quality model audioPath=\(audioPath) realtimePreviewChars=\(realtimeText?.count ?? 0)")
                 self.fallbackTranscribe(
                     audioPath: audioPath,
                     curMode: curMode,

package/src/native/Recordings/RecordingsTests/OpenAIAPIKeyStoreTests.swift

@@ -119,7 +119,7 @@ struct OpenAIAPIKeyStoreTests {
         let home = try makeHome()
         try writeConfig(home: home, [
             "openai_api_key": "old-key",
-            "transcription_model": "gpt-4o-mini-transcribe",
+            "transcription_model": "gpt-4o-transcribe",
         ])
 
         try OpenAIAPIKeyStore.save(key: "sk-rotated", homePath: home.path)
@@ -131,7 +131,7 @@ struct OpenAIAPIKeyStoreTests {
         let data = try Data(contentsOf: configURL)
         let json = try #require(JSONSerialization.jsonObject(with: data) as? [String: Any])
         #expect(json["openai_api_key"] as? String == "sk-rotated")
-        #expect(json["transcription_model"] as? String == "gpt-4o-mini-transcribe")
+        #expect(json["transcription_model"] as? String == "gpt-4o-transcribe")
     }
 
     @Test("Saving an empty key removes it from config.json")