v0.0.6 - Fix doubled transcript output by skipping autotype when the app window has focus

src/gui_main.c

@@ -11,6 +11,7 @@
 #include <stdlib.h>
 #include <string.h>
 #include <unistd.h>
+#include <time.h>
 
 typedef struct {
     GtkApplication *app;
@@ -42,6 +43,8 @@ typedef struct {
     unsigned int meter_timer_id;
     float live_peak;
     int runtime_ready;
+    char last_transcript[4096];
+    double last_transcript_time;
 } gui_state_t;
 
 static float clampf(float v, float lo, float hi) {
@@ -163,6 +166,28 @@ static int peak_is_speech(const float *samples, size_t count, float threshold) {
     return peak_abs >= threshold;
 }
 
+static double monotonic_seconds(void) {
+    struct timespec ts;
+    clock_gettime(CLOCK_MONOTONIC, &ts);
+    return (double)ts.tv_sec + (double)ts.tv_nsec / 1000000000.0;
+}
+
+static int is_duplicate_transcript(gui_state_t *s, const char *text) {
+    if (!s || !text || text[0] == '\0') return 0;
+    if (s->last_transcript[0] == '\0') return 0;
+
+    const double dt = monotonic_seconds() - s->last_transcript_time;
+    if (dt < 0.0 || dt > 1.0) return 0;
+
+    return strcmp(s->last_transcript, text) == 0;
+}
+
+static void remember_transcript(gui_state_t *s, const char *text) {
+    if (!s || !text || text[0] == '\0') return;
+    snprintf(s->last_transcript, sizeof(s->last_transcript), "%s", text);
+    s->last_transcript_time = monotonic_seconds();
+}
+
 static void process_segment(gui_state_t *s, float *samples, size_t count) {
     if (!samples || count == 0) {
         free(samples);
@@ -178,7 +203,13 @@ static void process_segment(gui_state_t *s, float *samples, size_t count) {
         return;
     }
 
-    if (gtk_toggle_button_get_active(GTK_TOGGLE_BUTTON(s->type_check))) {
+    int should_autotype = gtk_toggle_button_get_active(GTK_TOGGLE_BUTTON(s->type_check)) ? 1 : 0;
+    if (should_autotype && gtk_window_is_active(GTK_WINDOW(s->window))) {
+        should_autotype = 0;
+        fprintf(stderr, "gui: autotype skipped because voice_linux window is focused\n");
+    }
+
+    if (should_autotype) {
         if (typer_capture_focus() != 0) {
             fprintf(stderr, "gui: failed to capture focus target for autotype\n");
         }
@@ -200,9 +231,16 @@ static void process_segment(gui_state_t *s, float *samples, size_t count) {
         return;
     }
 
+    if (is_duplicate_transcript(s, text)) {
+        fprintf(stderr, "gui: suppressed duplicate transcript within 1s window\n");
+        set_status(s, "Duplicate segment suppressed");
+        free(text);
+        return;
+    }
+
     append_transcript(s, text);
 
-    if (gtk_toggle_button_get_active(GTK_TOGGLE_BUTTON(s->type_check))) {
+    if (should_autotype) {
         if (typer_type_text(text) != 0) {
             fprintf(stderr, "gui: autotype injection failed\n");
             set_status(s, "Autotype failed (see terminal logs)");
@@ -211,6 +249,7 @@ static void process_segment(gui_state_t *s, float *samples, size_t count) {
         }
     }
 
+    remember_transcript(s, text);
     set_status(s, audio_msg);
     free(text);
 }

src/version.h

@@ -1,9 +1,9 @@
 #ifndef VOICE_VERSION_H
 #define VOICE_VERSION_H
 
-#define VOICE_LINUX_VERSION "v0.0.5"
+#define VOICE_LINUX_VERSION "v0.0.6"
 #define VOICE_LINUX_VERSION_MAJOR 0
 #define VOICE_LINUX_VERSION_MINOR 0
-#define VOICE_LINUX_VERSION_PATCH 5
+#define VOICE_LINUX_VERSION_PATCH 6
 
 #endif