Update app.py
Browse filesFix for Safari microphone
app.py
CHANGED
@@ -83,7 +83,13 @@ def transcribe_audio_online_streaming(file, language):
|
|
83 |
while recognizer.is_ready(s):
|
84 |
recognizer.decode_streams([s])
|
85 |
|
86 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
87 |
|
88 |
except Exception as e:
|
89 |
yield f"Error: {e}"
|
@@ -112,6 +118,8 @@ def transcribe_microphone_stream(audio_chunk, stream_state, language):
|
|
112 |
return "", None
|
113 |
|
114 |
sample_rate, waveform_np = audio_chunk
|
|
|
|
|
115 |
|
116 |
# Resample if needed
|
117 |
if sample_rate != 16000:
|
@@ -133,6 +141,11 @@ def transcribe_microphone_stream(audio_chunk, stream_state, language):
|
|
133 |
recognizer.decode_streams([stream_state])
|
134 |
|
135 |
current_text = recognizer.get_result(stream_state)
|
|
|
|
|
|
|
|
|
|
|
136 |
|
137 |
return current_text, stream_state
|
138 |
|
|
|
83 |
while recognizer.is_ready(s):
|
84 |
recognizer.decode_streams([s])
|
85 |
|
86 |
+
current_text = recognizer.get_result(s)
|
87 |
+
if isinstance(current_text, (list, np.ndarray)):
|
88 |
+
current_text = " ".join(map(str, current_text))
|
89 |
+
elif isinstance(current_text, bytes):
|
90 |
+
current_text = current_text.decode("utf-8", errors="ignore")
|
91 |
+
|
92 |
+
yield current_text
|
93 |
|
94 |
except Exception as e:
|
95 |
yield f"Error: {e}"
|
|
|
118 |
return "", None
|
119 |
|
120 |
sample_rate, waveform_np = audio_chunk
|
121 |
+
if len(waveform_np.shape) > 1:
|
122 |
+
waveform_np = waveform_np.mean(axis=1)
|
123 |
|
124 |
# Resample if needed
|
125 |
if sample_rate != 16000:
|
|
|
141 |
recognizer.decode_streams([stream_state])
|
142 |
|
143 |
current_text = recognizer.get_result(stream_state)
|
144 |
+
|
145 |
+
if isinstance(current_text, (list, np.ndarray)):
|
146 |
+
current_text = " ".join(map(str, current_text))
|
147 |
+
elif isinstance(current_text, bytes):
|
148 |
+
current_text = current_text.decode("utf-8", errors="ignore")
|
149 |
|
150 |
return current_text, stream_state
|
151 |
|