Reduce model initialization time for offline speech recognition (#213)
This commit is contained in:
@@ -86,6 +86,7 @@ class OfflineRecognizer(object):
|
||||
num_threads=num_threads,
|
||||
debug=debug,
|
||||
provider=provider,
|
||||
model_type="transducer",
|
||||
)
|
||||
|
||||
feat_config = OfflineFeatureExtractorConfig(
|
||||
@@ -149,6 +150,7 @@ class OfflineRecognizer(object):
|
||||
num_threads=num_threads,
|
||||
debug=debug,
|
||||
provider=provider,
|
||||
model_type="paraformer",
|
||||
)
|
||||
|
||||
feat_config = OfflineFeatureExtractorConfig(
|
||||
@@ -211,6 +213,7 @@ class OfflineRecognizer(object):
|
||||
num_threads=num_threads,
|
||||
debug=debug,
|
||||
provider=provider,
|
||||
model_type="nemo_ctc",
|
||||
)
|
||||
|
||||
feat_config = OfflineFeatureExtractorConfig(
|
||||
|
||||
Reference in New Issue
Block a user