From 4405c451ce419539246b056112d44da70850c641 Mon Sep 17 00:00:00 2001
From: koiking213 <koiking213@gmail.com>
Date: Tue, 20 Feb 2024 23:29:25 +0900
Subject: [PATCH] specify dtype for librosa.load, instead of cast

---
 whisper_online.py        | 4 ++--
 whisper_online_server.py | 5 +++--
 2 files changed, 5 insertions(+), 4 deletions(-)

diff --git a/whisper_online.py b/whisper_online.py
index bcaeec4..d79396a 100644
--- a/whisper_online.py
+++ b/whisper_online.py
@@ -10,8 +10,8 @@ import math
 
 @lru_cache
 def load_audio(fname):
-    a, _ = librosa.load(fname, sr=16000)
-    return a.astype('float32')
+    a, _ = librosa.load(fname, sr=16000, dtype=np.float32)
+    return a
 
 def load_audio_chunk(fname, beg, end):
     audio = load_audio(fname)
diff --git a/whisper_online_server.py b/whisper_online_server.py
index a609ef0..0cdc97d 100644
--- a/whisper_online_server.py
+++ b/whisper_online_server.py
@@ -4,6 +4,7 @@ from whisper_online import *
 import sys
 import argparse
 import os
+import numpy as np
 parser = argparse.ArgumentParser()
 
 # server options
@@ -137,8 +138,8 @@ class ServerProcessor:
             if not raw_bytes:
                 break
             sf = soundfile.SoundFile(io.BytesIO(raw_bytes), channels=1,endian="LITTLE",samplerate=SAMPLING_RATE, subtype="PCM_16",format="RAW")
-            audio, _ = librosa.load(sf,sr=SAMPLING_RATE)
-            out.append(audio.astype('float32'))
+            audio, _ = librosa.load(sf,sr=SAMPLING_RATE,dtype=np.float32)
+            out.append(audio)
         if not out:
             return None
         return np.concatenate(out)