Skip to content

Python tool for converting files and office documents to Markdown.

Add language parameter for the audio transcription (#1295)

hchmch5610ced394b3
diff --git a/packages/markitdown/src/markitdown/converters/_audio_converter.py b/packages/markitdown/src/markitdown/converters/_audio_converter.py
index 3d96b53c..c4393e92 100644
--- a/packages/markitdown/src/markitdown/converters/_audio_converter.py
+++ b/packages/markitdown/src/markitdown/converters/_audio_converter.py
@@ -88,10 +88,15 @@ class AudioConverter(DocumentConverter):
else:
audio_format = None
+ # Figure out the audio language for transcription, default to "en-US"
+ language = kwargs.get("language", "en-US")
+
# Transcribe
if audio_format:
try:
- transcript = transcribe_audio(file_stream, audio_format=audio_format)
+ transcript = transcribe_audio(
+ file_stream, audio_format=audio_format, language=language
+ )
if transcript:
md_content += "\n\n### Audio Transcript:\n" + transcript
except MissingDependencyException:
diff --git a/packages/markitdown/src/markitdown/converters/_transcribe_audio.py b/packages/markitdown/src/markitdown/converters/_transcribe_audio.py
index 6d64d4ea..6caa2c65 100644
--- a/packages/markitdown/src/markitdown/converters/_transcribe_audio.py
+++ b/packages/markitdown/src/markitdown/converters/_transcribe_audio.py
@@ -26,7 +26,9 @@ except ImportError:
_dependency_exc_info = sys.exc_info()
-def transcribe_audio(file_stream: BinaryIO, *, audio_format: str = "wav") -> str:
+def transcribe_audio(
+ file_stream: BinaryIO, *, audio_format: str = "wav", language: str = "en-US"
+) -> str:
# Check for installed dependencies
if _dependency_exc_info is not None:
raise MissingDependencyException(
@@ -54,5 +56,5 @@ def transcribe_audio(file_stream: BinaryIO, *, audio_format: str = "wav") -> str
recognizer = sr.Recognizer()
with sr.AudioFile(audio_source) as source:
audio = recognizer.record(source)
- transcript = recognizer.recognize_google(audio).strip()
+ transcript = recognizer.recognize_google(audio, language=language).strip()
return "[No speech detected]" if transcript == "" else transcript