diff --git a/tts_engine.py b/tts_engine.py
index 27b6c4d..c817755 100644
--- a/tts_engine.py
+++ b/tts_engine.py
@@ -41,7 +41,6 @@ import struct
 import subprocess
 import wave
 import logging
-from bitarray import bitarray
 
 logger = logging.getLogger(__name__)
 
@@ -63,55 +62,12 @@ DV3K_AUDIO_FIELD_ID = 0x00
 
 DV3K_SAMPLES_PER_FRAME = 160
 
-DV3K_RATEP_DMR = bytes([
-    0x61, 0x00, 0x0D, 0x00, 0x0A,
-    0x01, 0x30, 0x07, 0x63, 0x40,
-    0x00, 0x00, 0x00, 0x00, 0x00,
-    0x00, 0x48
+DV3K_RATET_DMR = bytes([
+    0x61, 0x00, 0x02, 0x00, 0x09, 0x21
 ])
 
 DV3K_PRODID_REQ = bytes([0x61, 0x00, 0x01, 0x00, 0x30])
 
-AMBE_SILENCE = bytes([0xAC, 0xAA, 0x40, 0x20, 0x00, 0x44, 0x40, 0x80, 0x80])
-
-
-def _interleave_ambe_to_dmr(frames_data):
-    '''
-    Convert raw AMBE+2 frames (9 bytes/72 bits each, from DV3000) to DMR burst format.
-
-    DMR voice bursts carry 3 AMBE frames per burst (216 bits = 2 x 108 bits).
-    The 3 frames are interleaved: burst_payload[i*3 + j] = frame[j].bit[i]
-    where i = 0..71 (bit index within frame), j = 0..2 (frame index).
-
-    This matches ETSI TS 102 361-1 voice burst payload format.
-    The output is compatible with readAMBE.readSingleFile() which splits
-    the data into 108-bit pairs for mk_voice.pkt_gen().
-    '''
-    result = bitarray(endian='big')
-
-    for idx in range(0, len(frames_data), 3):
-        f0 = frames_data[idx] if idx < len(frames_data) else AMBE_SILENCE
-        f1 = frames_data[idx + 1] if idx + 1 < len(frames_data) else AMBE_SILENCE
-        f2 = frames_data[idx + 2] if idx + 2 < len(frames_data) else AMBE_SILENCE
-
-        b0 = bitarray(endian='big')
-        b0.frombytes(f0)
-        b1 = bitarray(endian='big')
-        b1.frombytes(f1)
-        b2 = bitarray(endian='big')
-        b2.frombytes(f2)
-
-        burst = bitarray(216, endian='big')
-        burst.setall(False)
-        for i in range(72):
-            burst[i * 3 + 0] = b0[i]
-            burst[i * 3 + 1] = b1[i]
-            burst[i * 3 + 2] = b2[i]
-
-        result.extend(burst)
-
-    return result.tobytes()
-
 
 def _get_tts_lang(announcement_language):
     if announcement_language in _LANG_MAP:
@@ -248,15 +204,15 @@ def _encode_ambe_ambeserver(wav_path, ambe_path, host, port):
         return False
 
     try:
-        sock.sendto(DV3K_RATEP_DMR, (host, port))
+        sock.sendto(DV3K_RATET_DMR, (host, port))
         data, addr = sock.recvfrom(1024)
         if data[0] != DV3K_START_BYTE:
-            logger.error('(TTS-AMBESERVER) Error configurando RATEP DMR')
+            logger.error('(TTS-AMBESERVER) Error configurando RATET DMR')
             sock.close()
             return False
-        logger.info('(TTS-AMBESERVER) RATEP DMR (AMBE+2) configurado')
+        logger.info('(TTS-AMBESERVER) RATET DMR (AMBE+2 tabla 33) configurado')
     except socket.timeout:
-        logger.error('(TTS-AMBESERVER) Timeout configurando RATEP')
+        logger.error('(TTS-AMBESERVER) Timeout configurando RATET')
         sock.close()
         return False
 
@@ -319,14 +275,9 @@ def _encode_ambe_ambeserver(wav_path, ambe_path, host, port):
         return False
 
     try:
-        interleaved = _interleave_ambe_to_dmr(_ambe_frames)
         with open(ambe_path, 'wb') as f:
-            f.write(interleaved)
-        _bursts = len(_ambe_frames) // 3
-        if len(_ambe_frames) % 3:
-            _bursts += 1
-        logger.info('(TTS-AMBESERVER) Entrelazado DMR: %d frames -> %d bursts (%d bytes)',
-                    len(_ambe_frames), _bursts, len(interleaved))
+            for frame in _ambe_frames:
+                f.write(frame)
     except Exception as e:
         logger.error('(TTS-AMBESERVER) Error escribiendo archivo AMBE: %s', e)
         return False