batchalign 0.7.5a2__tar.gz → 0.7.5a4__tar.gz

{batchalign-0.7.5a2/batchalign.egg-info → batchalign-0.7.5a4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: batchalign
-Version: 0.7.5a2
+Version: 0.7.5a4
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu

{batchalign-0.7.5a2 → batchalign-0.7.5a4}/batchalign/cli/cli.py

@@ -247,6 +247,12 @@ def utseg(ctx, in_dir, out_dir, lang, num_speakers, **kwargs):
 @common_options
 @click.option("--whisper/--rev",
               default=False, help="Use OpenAI Whisper (ASR) instead of Rev.AI (default).")
+@click.option("--lang",
+              help="sample language in three-letter ISO 3166-1 alpha-3 code",
+              show_default=True,
+              default="eng",
+              type=str)
+@click.option("-n", "--num_speakers", type=int, help="number of speakers in the language sample", default=2)
 @click.pass_context
 def benchmark(ctx, in_dir, out_dir, lang, num_speakers, whisper, **kwargs):
     """Benchmark ASR utilities for their word accuracy"""

{batchalign-0.7.5a2 → batchalign-0.7.5a4}/batchalign/pipelines/asr/rev.py

@@ -63,13 +63,16 @@ class RevEngine(BatchalignEngine):
 
         L.info(f"Uploading '{pathlib.Path(f).stem}'...")
         # we will send the file for processing
-        job = client.submit_job_local_file(f,
-                                           metadata=f"batchalign2_{pathlib.Path(f).stem}",
-                                           language=lang,
-                                           # some languages don't have postprocessors, so this option
-                                           # raises an exception
-                                           skip_postprocessing=(True if resolve("utterance", self.__lang_code) and lang in ["en", "fr"] else False),
-                                           speakers_count=self.__num_speakers)
+        if "en" not in lang and "es" not in lang:
+            job = client.submit_job_local_file(f,
+                                               metadata=f"batchalign2_{pathlib.Path(f).stem}",
+                                               language=lang)
+        else:
+            job = client.submit_job_local_file(f,
+                                               metadata=f"batchalign2_{pathlib.Path(f).stem}",
+                                               language=lang,
+                                               skip_postprocessing=(True if resolve("utterance", self.__lang_code) and lang in ["en", "fr"] else False),
+                                               speakers_count=self.__num_speakers)
 
         # we will wait untitl job finishes
         status = client.get_job_details(job.id).status

batchalign-0.7.5a4/batchalign/version

@@ -0,0 +1,3 @@
+0.7.5-alpha.4
+September 3nd, 2024
+fix benchmark command

{batchalign-0.7.5a2 → batchalign-0.7.5a4/batchalign.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: batchalign
-Version: 0.7.5a2
+Version: 0.7.5a4
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu

batchalign-0.7.5a2/batchalign/version

@@ -1,3 +0,0 @@
-0.7.5-alpha.2
-September 3nd, 2024
-Some tagging modification