diff --git a/src/benchmark/benchmark_generator.py b/src/benchmark/benchmark_generator.py
index a1ee1b9..4dcb388 100644
--- a/src/benchmark/benchmark_generator.py
+++ b/src/benchmark/benchmark_generator.py
@@ -78,7 +78,7 @@ def synthesize(model, phoneme_ids, speaker_id, sample_rate) -> float:
     )
     end_time = time.monotonic_ns()
 
-    audio_sec = (len(audio) / 2) / sample_rate
+    audio_sec = len(audio) / sample_rate
     infer_sec = (end_time - start_time) / 1e9
     rtf = infer_sec / audio_sec
 
diff --git a/src/benchmark/benchmark_onnx.py b/src/benchmark/benchmark_onnx.py
index 9281810..3fa0c2b 100644
--- a/src/benchmark/benchmark_onnx.py
+++ b/src/benchmark/benchmark_onnx.py
@@ -107,7 +107,7 @@ def synthesize(session, phoneme_ids, speaker_id, sample_rate) -> float:
     )[0].squeeze()
     end_time = time.monotonic_ns()
 
-    audio_sec = (len(audio) / 2) / sample_rate
+    audio_sec = len(audio) / sample_rate
     infer_sec = (end_time - start_time) / 1e9
     rtf = infer_sec / audio_sec
 
diff --git a/src/benchmark/benchmark_torchscript.py b/src/benchmark/benchmark_torchscript.py
index f1972df..d4bee21 100644
--- a/src/benchmark/benchmark_torchscript.py
+++ b/src/benchmark/benchmark_torchscript.py
@@ -1,5 +1,6 @@
 #!/usr/bin/env python3
 import argparse
+import logging
 import json
 import time
 import statistics
@@ -11,6 +12,8 @@ _NOISE_SCALE = 0.667
 _LENGTH_SCALE = 1.0
 _NOISE_W = 0.8
 
+_LOGGER = logging.getLogger(__name__)
+
 
 def main() -> None:
     parser = argparse.ArgumentParser()
@@ -19,6 +22,7 @@ def main() -> None:
     )
     parser.add_argument("-c", "--config", help="Path to model config file (.json)")
     args = parser.parse_args()
+    logging.basicConfig(level=logging.DEBUG)
 
     if not args.config:
         args.config = f"{args.model}.json"
@@ -81,10 +85,18 @@ def synthesize(model, phoneme_ids, speaker_id, sample_rate) -> float:
     )
     end_time = time.monotonic_ns()
 
-    audio_sec = (len(audio) / 2) / sample_rate
+    audio_sec = len(audio) / sample_rate
     infer_sec = (end_time - start_time) / 1e9
+    rtf = infer_sec / audio_sec
 
-    return infer_sec / audio_sec
+    _LOGGER.debug(
+        "Real-time factor: %s (infer=%s sec, audio=%s sec)",
+        rtf,
+        infer_sec,
+        audio_sec,
+    )
+
+    return rtf
 
 
 if __name__ == "__main__":