Update handler.py
Browse files- handler.py +13 -51
handler.py
CHANGED
@@ -4,27 +4,31 @@ import os
|
|
4 |
from io import StringIO
|
5 |
from typing import Dict, Any
|
6 |
|
|
|
7 |
from transformers import pipeline
|
8 |
|
9 |
|
10 |
class EndpointHandler:
|
11 |
|
12 |
def __init__(self, asr_model_path: str = "./whisper-large-v2"):
|
|
|
|
|
|
|
13 |
# Create an ASR pipeline using the model located in the specified directory
|
14 |
self.asr_pipeline = pipeline(
|
15 |
"automatic-speech-recognition",
|
16 |
model = asr_model_path,
|
|
|
17 |
)
|
18 |
|
19 |
def __call__(self, data: Dict[str, Any]) -> str:
|
20 |
|
21 |
-
|
22 |
-
if "audio_data" not in json_data.keys():
|
23 |
raise Exception("Request must contain a top-level key named 'audio_data'")
|
24 |
|
25 |
# Get the audio data from the input
|
26 |
-
audio_data =
|
27 |
-
|
28 |
|
29 |
# Decode the binary audio data if it's provided as a base64 string
|
30 |
if isinstance(audio_data, str):
|
@@ -33,12 +37,11 @@ class EndpointHandler:
|
|
33 |
# Process the audio data with the ASR pipeline
|
34 |
transcription = self.asr_pipeline(
|
35 |
audio_data,
|
36 |
-
return_timestamps=
|
37 |
-
chunk_length_s=30,
|
38 |
-
batch_size=8,
|
39 |
-
|
40 |
-
|
41 |
-
generate_kwargs={"task": "transcribe", "language": "<|language|>"}
|
42 |
)
|
43 |
|
44 |
# Convert the transcription to JSON
|
@@ -46,44 +49,3 @@ class EndpointHandler:
|
|
46 |
json.dump(transcription, result)
|
47 |
|
48 |
return result.getvalue()
|
49 |
-
|
50 |
-
def init():
|
51 |
-
global asr_pipeline
|
52 |
-
# Set the path to the directory where the model is stored
|
53 |
-
model_path = os.getenv("AZUREML_MODEL_DIR", "./whisper-large-v2")
|
54 |
-
|
55 |
-
# Create an ASR pipeline using the model located in the specified directory
|
56 |
-
asr_pipeline = pipeline(
|
57 |
-
"automatic-speech-recognition",
|
58 |
-
model = model_path,
|
59 |
-
)
|
60 |
-
|
61 |
-
|
62 |
-
def run(raw_data):
|
63 |
-
json_data = json.loads(raw_data)
|
64 |
-
if "audio_data" not in json_data.keys():
|
65 |
-
raise Exception("Request must contain a top level key named 'audio_data'")
|
66 |
-
|
67 |
-
# Get the audio data from the input
|
68 |
-
audio_data = json_data["audio_data"]
|
69 |
-
|
70 |
-
# Decode the binary audio data if it's provided as a base64 string
|
71 |
-
if isinstance(audio_data, str):
|
72 |
-
import base64
|
73 |
-
audio_data = base64.b64decode(audio_data)
|
74 |
-
|
75 |
-
# Process the audio data with the ASR pipeline
|
76 |
-
transcription = asr_pipeline(
|
77 |
-
audio_data,
|
78 |
-
return_timestamps = False,
|
79 |
-
chunk_length_s = 30,
|
80 |
-
batch_size = 8,
|
81 |
-
max_new_tokens = 1000,
|
82 |
-
generate_kwargs = {"task": "transcribe", "language": "<|de|>"}
|
83 |
-
)
|
84 |
-
|
85 |
-
# Convert the transcription to JSON
|
86 |
-
result = StringIO()
|
87 |
-
json.dump(transcription, result)
|
88 |
-
|
89 |
-
return result.getvalue()
|
|
|
4 |
from io import StringIO
|
5 |
from typing import Dict, Any
|
6 |
|
7 |
+
import torch
|
8 |
from transformers import pipeline
|
9 |
|
10 |
|
11 |
class EndpointHandler:
|
12 |
|
13 |
def __init__(self, asr_model_path: str = "./whisper-large-v2"):
|
14 |
+
device = 0 if torch.cuda.is_available() else -1
|
15 |
+
device = -1
|
16 |
+
print("Using device:", device)
|
17 |
# Create an ASR pipeline using the model located in the specified directory
|
18 |
self.asr_pipeline = pipeline(
|
19 |
"automatic-speech-recognition",
|
20 |
model = asr_model_path,
|
21 |
+
device = device
|
22 |
)
|
23 |
|
24 |
def __call__(self, data: Dict[str, Any]) -> str:
|
25 |
|
26 |
+
if "audio_data" not in data.keys():
|
|
|
27 |
raise Exception("Request must contain a top-level key named 'audio_data'")
|
28 |
|
29 |
# Get the audio data from the input
|
30 |
+
audio_data = data["audio_data"]
|
31 |
+
options = data["options"]
|
32 |
|
33 |
# Decode the binary audio data if it's provided as a base64 string
|
34 |
if isinstance(audio_data, str):
|
|
|
37 |
# Process the audio data with the ASR pipeline
|
38 |
transcription = self.asr_pipeline(
|
39 |
audio_data,
|
40 |
+
return_timestamps = True,
|
41 |
+
chunk_length_s = 30,
|
42 |
+
batch_size = 8,
|
43 |
+
max_new_tokens = 10000,
|
44 |
+
generate_kwargs = options
|
|
|
45 |
)
|
46 |
|
47 |
# Convert the transcription to JSON
|
|
|
49 |
json.dump(transcription, result)
|
50 |
|
51 |
return result.getvalue()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|