draft: update clients for separate utterances (#38)

nvidia-riva · Mar 23, 2023 · e7b4c61 · e7b4c61
1 parent db77d10
commit e7b4c61
Show file tree

Hide file tree

Showing 3 changed files with 7 additions and 4 deletions.
diff --git a/riva/client/asr.py b/riva/client/asr.py
@@ -253,7 +253,10 @@ def print_streaming(
 def print_offline(response: rasr.RecognizeResponse) -> None:
     print(response)
     if len(response.results) > 0 and len(response.results[0].alternatives) > 0:
-        print("Final transcript:", response.results[0].alternatives[0].transcript)
+        final_transcript = ""
+        for res in response.results:
+            final_transcript += res.alternatives[0].transcript
+        print("Final transcript:", final_transcript)
 
 
 def streaming_request_generator(

diff --git a/scripts/asr/transcribe_file.py b/scripts/asr/transcribe_file.py
@@ -49,7 +49,7 @@ def parse_args() -> argparse.Namespace:
         "--print-confidence", action="store_true", help="Whether to print stability and confidence of transcript."
     )
     parser = add_connection_argparse_parameters(parser)
-    parser = add_asr_config_argparse_parameters(parser, profanity_filter=True)
+    parser = add_asr_config_argparse_parameters(parser, max_alternatives=True, profanity_filter=True, word_time_offsets=True)
     args = parser.parse_args()
     if not args.list_devices and args.input_file is None:
         parser.error(

diff --git a/scripts/asr/transcribe_file_offline.py b/scripts/asr/transcribe_file_offline.py
@@ -18,7 +18,7 @@ def parse_args() -> argparse.Namespace:
     )
     parser.add_argument("--input-file", required=True, type=Path, help="A path to a local file to transcribe.")
     parser = add_connection_argparse_parameters(parser)
-    parser = add_asr_config_argparse_parameters(parser, profanity_filter=True, word_time_offsets=True)
+    parser = add_asr_config_argparse_parameters(parser, max_alternatives=True, profanity_filter=True, word_time_offsets=True)
     args = parser.parse_args()
     args.input_file = args.input_file.expanduser()
     return args
@@ -31,7 +31,7 @@ def main() -> None:
     config = riva.client.RecognitionConfig(
         encoding=riva.client.AudioEncoding.LINEAR_PCM,
         language_code=args.language_code,
-        max_alternatives=1,
+        max_alternatives=args.max_alternatives,
         profanity_filter=args.profanity_filter,
         enable_automatic_punctuation=args.automatic_punctuation,
         verbatim_transcripts=not args.no_verbatim_transcripts,