fix RTP packet pacing

agent_call/real_cell.py

@@ -340,6 +340,7 @@ class RTPMedia:
         self.sequence = 0
         self.timestamp = 0
         self.ssrc = uuid.uuid4().int & 0xFFFFFFFF
+        self._next_send_at: float | None = None
         self._send_lock = threading.Lock()
 
     @property
@@ -399,11 +400,18 @@ class RTPMedia:
             )
         payload = pcm16_to_alaw(raw, input_rate_hz, 8000)
         sent = 0
+        packet_interval_s = 0.02
         with self._send_lock:
+            now = time.monotonic()
+            if self._next_send_at is None or self._next_send_at < now - packet_interval_s:
+                self._next_send_at = now
             for offset in range(0, len(payload), 160):
                 frame = payload[offset : offset + 160]
                 if not frame:
                     continue
+                delay = self._next_send_at - time.monotonic()
+                if delay > 0:
+                    time.sleep(delay)
                 header = struct.pack(
                     "!BBHII",
                     0x80,
@@ -421,7 +429,9 @@ class RTPMedia:
                 self.sequence += 1
                 self.timestamp += 160
                 sent += len(frame)
-                time.sleep(0.02)
+                self._next_send_at = max(
+                    self._next_send_at + packet_interval_s, time.monotonic()
+                )
         return sent
 
     @staticmethod

tests/test_real_cell.py

@@ -10,6 +10,7 @@ import wave
 from pathlib import Path
 from types import SimpleNamespace
 from typing import Any, cast
+from unittest.mock import patch
 
 from agent_call.real_cell import (
     CellCallConfig,
@@ -154,6 +155,35 @@ class RealCellTests(unittest.TestCase):
         self.assertEqual(fake_ari.record_requests, 0)
         self.assertNotIn("opening", order)
 
+    def test_rtp_send_keeps_packet_clock_across_frames(self) -> None:
+        media = RTPMedia("127.0.0.1", 0)
+        receiver = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
+        receiver.bind(("127.0.0.1", 0))
+        media.peer = ("127.0.0.1", receiver.getsockname()[1])
+        clock = [0.0]
+        sleeps: list[float] = []
+
+        def monotonic() -> float:
+            return clock[0]
+
+        def sleep(seconds: float) -> None:
+            sleeps.append(seconds)
+            clock[0] += seconds
+
+        try:
+            with (
+                patch("agent_call.real_cell.time.monotonic", monotonic),
+                patch("agent_call.real_cell.time.sleep", sleep),
+            ):
+                media.send_pcm16(bytes([0, 0]) * 160, 8000)
+                media.send_pcm16(bytes([0, 0]) * 160, 8000)
+            self.assertEqual(sleeps, [0.02])
+            self.assertEqual(len(receiver.recvfrom(512)[0]), 172)
+            self.assertEqual(len(receiver.recvfrom(512)[0]), 172)
+        finally:
+            media.close()
+            receiver.close()
+
     def test_pcma_decode_and_rtp_payload(self) -> None:
         media = RTPMedia("127.0.0.1", 0)
         receiver = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)