BatchalignHK 0.7.22.post25__tar.gz → 0.7.22.post27__tar.gz

{batchalignhk-0.7.22.post25 → batchalignhk-0.7.22.post27}/BatchalignHK.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: BatchalignHK
-Version: 0.7.22.post25
+Version: 0.7.22.post27
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu

{batchalignhk-0.7.22.post25 → batchalignhk-0.7.22.post27}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: BatchalignHK
-Version: 0.7.22.post25
+Version: 0.7.22.post27
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu

{batchalignhk-0.7.22.post25 → batchalignhk-0.7.22.post27}/batchalign/formats/chat/generator.py

@@ -33,6 +33,10 @@ def generate_chat_utterance(utterance: Utterance, special_mor=False, write_wor=T
     if merge_letters:
         main_line = re.sub(r"([a-z]) ", r"\1", main_line)
         main_line = re.sub(r"([a-z])([^a-z])", r"\1 \2", main_line)
+        main_line = re.sub(r"([a-z]) (\d+)", r"\1\2", main_line)
+        main_line = re.sub(r"([a-z]) @", r"\1@", main_line)
+        main_line = re.sub(r"([a-z]) :", r"\1:", main_line)
+        main_line = re.sub(r"([a-z]) >", r"\1>", main_line)
     main_line = re.sub(r"«", "“", main_line)
     main_line = re.sub(r"»", "”", main_line)
     main_line = re.sub(r"—", "-", main_line)

{batchalignhk-0.7.22.post25 → batchalignhk-0.7.22.post27}/batchalign/pipelines/utr/tencent_utr.py

@@ -32,9 +32,13 @@ import base64
 from tencentcloud.common.credential import Credential
 from tencentcloud.asr.v20190614.asr_client import AsrClient, models
 
+from qcloud_cos import CosConfig
+from qcloud_cos import CosS3Client
+
 import asyncio
 import tempfile
 import os
+import uuid
 
 
 import logging
@@ -50,9 +54,21 @@ class TencentUTREngine(BatchalignEngine):
             try:
                 id = config["asr"]["engine.tencent.id"] 
                 key = config["asr"]["engine.tencent.key"] 
+                region = config["asr"]["engine.tencent.region"]
+                bucket_name = config["asr"]["engine.tencent.bucket"]
             except KeyError:
                 raise ConfigError("No Tencent Cloud key found. Tencent Cloud was not set up! Please write one yourself and place it at ~/.batchalign.ini.")
 
+        config = CosConfig(
+            Region=region,
+            SecretId=id,
+            SecretKey=key,
+            Token=None,
+            Scheme="https"
+        )
+        self.__bucket = CosS3Client(config)
+        self.__bucket_name = bucket_name
+
         self.__lang_code = lang
 
         if lang == "yue":
@@ -130,10 +146,20 @@ class TencentUTREngine(BatchalignEngine):
         # audio = AudioSegment.from_file(processed_path)
         
         L.info(f"Uploading '{pathlib.Path(f).stem}'...")
-        # we will send the file for processing
-        if not str(f).startswith("http"):
-            with open(f, "rb") as image_file:
-                encoded_string = base64.b64encode(image_file.read())
+        bucket = self.__bucket
+        bucket_name = self.__bucket_name
+        uid = str(uuid.uuid4())
+
+        response = bucket.upload_file(
+            Bucket=bucket_name,
+            LocalFilePath=f,
+            Key=uid+pathlib.Path(f).suffix,
+            PartSize=1,
+            MAXThread=10,
+            EnableMD5=False
+        )
+
+
 
         req = models.CreateRecTaskRequest()
         if lang in {'zho', 'yue', 'wuu', 'nan','hak'}:
@@ -143,12 +169,8 @@ class TencentUTREngine(BatchalignEngine):
         req.ResTextFormat = 1
         req.SpeakerDiarization = 1
         req.ChannelNum = 1
-        if not str(f).startswith("http"):
-            req.Data = encoded_string.decode('ascii')
-            req.SourceType = 1
-        else:
-            req.Url = f
-            req.SourceType = 0
+        req.Url = response["Location"]
+        req.SourceType = 0
         resp = client.CreateRecTask(req)
 
         L.info(f"Tencent is transcribing '{pathlib.Path(f).stem}'...")

batchalignhk-0.7.22.post27/batchalign/version

@@ -0,0 +1,3 @@
+0.7.22-post.27
+November 28rd, 2025
+Tencent Fixes

batchalignhk-0.7.22.post25/batchalign/version

@@ -1,3 +0,0 @@
-0.7.22-post.25
-November 23rd, 2025
-Remove bullets