plip_voice/main/conversation.c

@@ -84,12 +84,16 @@ static char s_sid[32] = {0};
  * NOT re-read dialer_current() for the greeting/reply — that polluted number
  * would 404 at the gateway. Capture the clean routed number here once. */
 static char s_number[16] = {0};
+/* Scene this call hints on (e.g. "SCENE_WARNING"), locked at pickup like the
+ * number. Empty for player-dialled calls. Drives the local SD hint clip. */
+static char s_scene[40] = {0};
 
 /* Incoming-call mode: an NPC "calls" the phone. conversation_arm_incoming()
- * stores the caller's persona number and rings; when the handset is then picked
- * up from IDLE, we skip the outgoing dialtone/dial/ringback and go straight to
- * GREET with this number (the NPC speaks first). */
+ * stores the caller's persona number (and optional scene) and rings; when the
+ * handset is then picked up from IDLE, we skip the outgoing dialtone/dial/
+ * ringback and go straight to GREET with this number (the NPC speaks first). */
 static char s_incoming_number[16] = {0};
+static char s_incoming_scene[40] = {0};
 static volatile bool s_incoming_armed = false;
 
 /* Known numbers: ringback when dialed */
@@ -105,6 +109,20 @@ static bool is_known(const char *num)
     return false;
 }
 
+/* "SCENE_WARNING" → "warning": strip the SCENE_ prefix and lowercase, matching
+ * the SD pack filenames from tools/tts/generate_plip_sd_pack.py. */
+static void scene_slug(const char *scene, char *out, size_t cap)
+{
+    const char *p = scene;
+    if (strncmp(p, "SCENE_", 6) == 0) p += 6;
+    size_t i = 0;
+    for (; p[i] && i + 1 < cap; i++) {
+        char c = p[i];
+        out[i] = (c >= 'A' && c <= 'Z') ? (char)(c - 'A' + 'a') : c;
+    }
+    out[i] = '\0';
+}
+
 static void go_idle(void)
 {
     tones_stop();
@@ -114,6 +132,8 @@ static void go_idle(void)
 #if CONFIG_PLIP_DIAL_DTMF
     dtmf_stop();
 #endif
+    s_scene[0] = '\0';   /* clear scripted-call scene so a later dial-out call
+                          * doesn't inherit it (CONNECTED would skip listening). */
     s_state = STATE_IDLE;
     ESP_LOGI(TAG, "-> IDLE");
 }
@@ -134,6 +154,7 @@ static void enter_incoming_greet(void)
     dialer_reset();
     s_offhook = true;
     snprintf(s_number, sizeof(s_number), "%s", s_incoming_number);
+    snprintf(s_scene, sizeof(s_scene), "%s", s_incoming_scene);
     snprintf(s_sid, sizeof(s_sid), "%lld", (long long)esp_timer_get_time());
     audio_pa_set(true);
     s_state = STATE_GREET;
@@ -277,11 +298,37 @@ static void conv_task(void *arg)
                     ESP_LOGW(TAG, "no SD clip + gateway greeting failed — silent");
                 }
             }
+            /* Scripted offline call: chain this scene's hint from the SD pack
+             * right after the greeting (the audio queue plays them back-to-back).
+             * Fully local — no gateway, no model. */
+            if (s_scene[0] != '\0' && audio_ensure_sd()) {
+                char slug[40], hint[80];
+                scene_slug(s_scene, slug, sizeof(slug));
+                snprintf(hint, sizeof(hint),
+                         "/sdcard/voice/hint_%s_l1_0.wav", slug);
+                FILE *hf = fopen(hint, "rb");
+                if (hf != NULL) {
+                    fclose(hf);
+                    audio_play_async(hint);
+                    ESP_LOGI(TAG, "GREET: chained SD hint %s", hint);
+                } else {
+                    ESP_LOGI(TAG, "GREET: no SD hint for scene %s (%s)",
+                             s_scene, hint);
+                }
+            }
             s_state = STATE_CONNECTED;
             ESP_LOGI(TAG, "-> CONNECTED");
             break;
 
         case STATE_CONNECTED:
+            /* Scripted offline call (greeting + SD hint already queued): there's
+             * no live two-way conversation to run — just let the clips play out
+             * and wait for the player to hang up. Skips the gateway/model listen
+             * loop entirely. The global on-hook check returns us to IDLE. */
+            if (s_scene[0] != '\0') {
+                vTaskDelay(pdMS_TO_TICKS(REPLY_POLL_MS));
+                break;
+            }
 #if CONFIG_PLIP_VOICE_REPLY
             /*
              * Stage 3 — LISTEN loop.
@@ -417,12 +464,14 @@ void conversation_on_hook_change(bool offhook)
     s_hook_changed = true;
 }
 
-void conversation_arm_incoming(const char *number)
+void conversation_arm_incoming(const char *number, const char *scene)
 {
     snprintf(s_incoming_number, sizeof(s_incoming_number), "%s",
              (number && number[0]) ? number : "17");
+    snprintf(s_incoming_scene, sizeof(s_incoming_scene), "%s",
+             (scene && scene[0]) ? scene : "");
     s_incoming_armed = true;
     phone_ring_start();
-    ESP_LOGI(TAG, "incoming call armed (num=%s) — ringing until pickup",
-             s_incoming_number);
+    ESP_LOGI(TAG, "incoming call armed (num=%s scene=%s) — ringing until pickup",
+             s_incoming_number, s_incoming_scene[0] ? s_incoming_scene : "-");
 }

plip_voice/main/conversation.h

@@ -6,5 +6,8 @@ void conversation_on_hook_change(bool offhook);
 
 /* Arm an INCOMING call from NPC persona `number` and start ringing. When the
  * handset is picked up from idle, the conversation jumps straight to the
- * greeting (persona speaks first) instead of the outgoing dial-tone path. */
-void conversation_arm_incoming(const char *number);
+ * greeting (persona speaks first) instead of the outgoing dial-tone path.
+ * `scene` (optional, e.g. "SCENE_WARNING") makes the phone play that scene's
+ * hint from its local SD pack right after the greeting; pass NULL/"" for a
+ * greeting-only call. */
+void conversation_arm_incoming(const char *number, const char *scene);

plip_voice/main/net.c

@@ -510,10 +510,13 @@ static esp_err_t handle_debug_ring(httpd_req_t *req)
 {
     /* /debug/ring[?number=NN] — arm an INCOMING call from persona NN (default
      * 17) and ring. On pickup the conversation jumps straight to the greeting. */
-    char query[48] = {0}, number[16] = {0};
+    char query[96] = {0}, number[16] = {0}, scene[40] = {0};
     httpd_req_get_url_query_str(req, query, sizeof(query));
     httpd_query_key_value(query, "number", number, sizeof(number));
-    conversation_arm_incoming(number);  /* defaults to "17" if empty; starts ringing */
+    httpd_query_key_value(query, "scene", scene, sizeof(scene));
+    /* scene (optional) lets the phone play that scene's hint from its SD pack
+     * after the greeting — fully offline, no model. */
+    conversation_arm_incoming(number, scene[0] ? scene : NULL);
     char resp[64];
     snprintf(resp, sizeof(resp), "{\"ok\":true,\"ringing\":true,\"incoming\":\"%s\"}",
              number[0] ? number : "17");