Remove references to transcription code from audio and video (#9)

* Remove references to transcription code from audio and video

* Add ARM build

Author: catileptic

.github/workflows/build.yml

@@ -76,7 +76,7 @@ jobs:
         if: (github.ref == 'refs/heads/main' || startsWith(github.ref, 'refs/tags')) && github.actor != 'dependabot[bot]'
         with:
           context: .
-          platforms: linux/amd64
+          platforms: linux/amd64, linux/arm64
           push: true
           tags: ${{ steps.meta.outputs.tags }}
           labels: ${{ steps.meta.outputs.labels }}

ingestors/media/audio.py

@@ -1,18 +1,15 @@
 import logging
-from datetime import datetime
 from followthemoney import model
 from pymediainfo import MediaInfo
-from normality import stringify
 
 from ingestors.ingestor import Ingestor
 from ingestors.support.timestamp import TimestampSupport
 from ingestors.exc import ProcessingException
-from ingestors.support.transcription import TranscriptionSupport
 
 log = logging.getLogger(__name__)
 
 
-class AudioIngestor(Ingestor, TimestampSupport, TranscriptionSupport):
+class AudioIngestor(Ingestor, TimestampSupport):
     MIME_TYPES = [
         "audio/mpeg",
         "audio/mp3",
@@ -59,22 +56,6 @@ def ingest(self, file_path, entity):
                 entity.add("duration", track.duration)
         except Exception as ex:
             raise ProcessingException(f"Could not read audio: {ex}") from ex
-        try:
-            start = datetime.now()
-            log.info(f"Attempting to transcribe {file_path}")
-            self.transcribe(file_path, entity)
-            elapsed_time = datetime.now() - start
-            # caution! this can't store an elapsed time larger than 24h
-            # datetime.seconds capped at [0,86400)
-            elapsed_time = divmod(elapsed_time.total_seconds(), 60)[0]
-            log.info(
-                f"Transcription duration: {elapsed_time} minutes (audio duration: {entity.get('duration')})"
-            )
-        except Exception as ex:
-            # If the transcription fails, the file processing should still count as a success.
-            # The existance of a transcription is not mandatory, for now.
-            entity.set("processingError", stringify(ex))
-            log.error(ex)
 
     @classmethod
     def match(cls, file_path, entity):

ingestors/media/video.py

@@ -1,18 +1,15 @@
 import logging
-from datetime import datetime
 from followthemoney import model
 from pymediainfo import MediaInfo
-from normality import stringify
 
 from ingestors.ingestor import Ingestor
 from ingestors.support.timestamp import TimestampSupport
 from ingestors.exc import ProcessingException
-from ingestors.support.transcription import TranscriptionSupport
 
 log = logging.getLogger(__name__)
 
 
-class VideoIngestor(Ingestor, TimestampSupport, TranscriptionSupport):
+class VideoIngestor(Ingestor, TimestampSupport):
     MIME_TYPES = [
         "application/x-shockwave-flash",
         "video/quicktime",
@@ -47,23 +44,6 @@ def ingest(self, file_path, entity):
                 entity.add("duration", track.duration)
         except Exception as ex:
             raise ProcessingException("Could not read video: %r", ex) from ex
-        try:
-            start = datetime.now()
-            log.info(f"Attempting to transcribe {file_path}")
-            audio_only_file = self.extract_audio(file_path)
-            self.transcribe(audio_only_file, entity)
-            elapsed_time = datetime.now() - start
-            # caution! this can't store an elapsed time larger than 24h
-            # datetime.seconds capped at [0,86400)
-            elapsed_time = divmod(elapsed_time.total_seconds(), 60)[0]
-            log.info(
-                f"Transcription duration: {elapsed_time} minutes (audio duration: {entity.get('duration')})"
-            )
-        except Exception as ex:
-            # If the transcription fails, the file processing should still count as a success.
-            # The existance of a transcription is not mandatory, for now.
-            entity.set("processingError", stringify(ex))
-            log.error(ex)
 
     @classmethod
     def match(cls, file_path, entity):