Fix transcriber bugs while processing folder

python/setup.py

@@ -68,7 +68,7 @@ def get_tag(self):
     cmdclass=cmdclass,
     python_requires='>=3',
     zip_safe=False, # Since we load so file from the filesystem, we can not run from zip file
-    setup_requires=['cffi>=1.0', 'requests', 'tqdm'],
-    install_requires=['cffi>=1.0', 'requests', 'tqdm'],
+    setup_requires=['cffi>=1.0', 'requests', 'tqdm', 'srt'],
+    install_requires=['cffi>=1.0', 'requests', 'tqdm', 'srt'],
     cffi_modules=['vosk_builder.py:ffibuilder'],
 )

python/vosk/transcriber/cli.py

@@ -56,22 +56,18 @@ def main():
         exit(1)
 
     if not Path(args.input).exists():
-        logging.info('File %s does not exist, please specify an existing file/directory' % (args.input))
-        exit(1)
-
-    if args.output !='' and not Path(args.output).exists():
-        logging.info('Output %s does not exist, please specify an existing file' % (args.output))
+        logging.info("File/folder '%s' does not exist, please specify an existing file/directory" % (args.input))
         exit(1)
 
     transcriber = Transcriber(args)
 
-    if Path(args.input).is_dir() and Path(args.output).is_dir():
+    if Path(args.input).is_dir():
         transcriber.process_dir(args)
         return
-    elif Path(args.input).is_file() and (args.output=='' or Path(args.output).is_file()):
+    elif Path(args.input).is_file():
         transcriber.process_file(args)
     else:
-        logging.info('Wrong arguments, input and output must be same type')
+        logging.info('Wrong arguments')
         exit(1)
 
 if __name__ == "__main__":

python/vosk/transcriber/transcriber.py

@@ -8,6 +8,7 @@
 from pathlib import Path
 from timeit import default_timer as timer
 from vosk import KaldiRecognizer, Model
+from multiprocessing.dummy import Pool
 
 class Transcriber:
 
@@ -76,11 +77,10 @@ def process_entry(self, inputdata):
             print(final_result)
         return final_result, tot_samples
 
-
-    def process_directory(self,args):
+    def process_dir(self,args):
         task_list = [(Path(args.input, fn), Path(args.output, Path(fn).stem).with_suffix('.' + args.output_type)) for fn in os.listdir(args.input)]
         with Pool() as pool:
-            pool.map(self.process_entry, file_list)
+            pool.map(self.process_entry, task_list)
 
     def process_file(self, args):
         start_time = timer()

travis/build-wheels-dockcross.sh

@@ -26,5 +26,5 @@ mkdir -p /io/wheelhouse/vosk-linux-$VOSK_ARCHITECTURE
 cp /opt/vosk-api/src/*.so /opt/vosk-api/src/vosk_api.h /io/wheelhouse/vosk-linux-$VOSK_ARCHITECTURE
 
 # Build wheel
-python3 -m pip install requests tqdm
+python3 -m pip install requests tqdm srt
 python3 -m pip wheel /opt/vosk-api/python --no-deps -w /io/wheelhouse