/* Extract the audio from mp4 files in subfolders of specified folder.
* In the "Fix" folder for a recording, there will be a subfolder for each
* segment of the recording: 00-03-00, 00-06-00, 00=09-00, etc.
* Each of these subfolders is initialized with three files:
* "ToFix.mp4" - the video of this segment
* "ToFix.flac" - the audio of this segment
* "ToFix.json" - the transcription of this segment
*/
public void ExtractAll(string inputFolder)
{
foreach (string dir in Directory.GetDirectories(inputFolder))
{
string inputFile = Path.Combine(dir, "ToFix.mp4");
// TODO - convert to mp3 instead of flac.
string outputFile = Path.Combine(dir, "ToFix.flac");
audioProcessing.Extract(inputFile, outputFile);
}
}
static void TranscribeVideo(
SampleVideo sample, // sample video to use
string fixedTags, // file in which to save the fixed transcription
string audio, // file in which to save the extracted audio
bool useSmallSample, // if true, use a small sample of the video/audio
bool useAudioFileAlreadyInCloud, // if true, use prior audio in cloud if it exists
string rawTranscription) // file in which to save the raw transcription
{
string videofilePath = sample.filepath;
string objectName = sample.objectname;
RepeatedField <string> phrases = sample.phrases;
AudioProcessing audioProcessing = new AudioProcessing();
string googleCloudBucketName = "govmeeting-transcribe";
TranscribeParameters transParams = new TranscribeParameters
{
audiofilePath = audio,
objectName = objectName,
GoogleCloudBucketName = googleCloudBucketName,
useAudioFileAlreadyInCloud = useAudioFileAlreadyInCloud,
language = "en",
MinSpeakerCount = 2,
MaxSpeakerCount = 6,
phrases = phrases
};
// Clean up from last run
File.Delete(audio);
File.Delete(fixedTags);
if (useSmallSample)
{
string shortVideoFile = videofilePath.Replace(".mp4", "-3min.mp4");
//SplitRecording splitRecording = new SplitRecording();
audioProcessing.ExtractPart(videofilePath, shortVideoFile, 60, 3 * 60);
videofilePath = shortVideoFile;
}
audioProcessing.Extract(videofilePath, audio);
GMFileAccess.SetGoogleCredentialsEnvironmentVariable();
// Transcribe the audio file
TranscribeAudio transcribe = new TranscribeAudio();
Transcribed_Dto response = transcribe.TranscribeAudioFile(transParams, rawTranscription);
string responseString = JsonConvert.SerializeObject(response, Formatting.Indented);
File.WriteAllText(fixedTags, responseString);
WriteCopyOfResponse(responseString, fixedTags);
}
public void Process(string videoFile, string meetingFolder, string language)
{
/////// Copy video to meeting folder /////////
AudioProcessing audioProcessing = new AudioProcessing();
string videofileCopy = Path.Combine(meetingFolder, "video.mp4");
// #### If MaxRecordingSize is not zero, we shorted the recording. ####
if (config.MaxRecordingSize == 0)
{
File.Copy(videoFile, videofileCopy);
}
else
{
audioProcessing.ExtractPart(videoFile, videofileCopy, 0, config.MaxRecordingSize);
}
/////// Extract the audio. ////////////////////////
ExtractAudio extract = new ExtractAudio();
string audioFile = Path.Combine(meetingFolder, "audio.flac");
audioProcessing.Extract(videofileCopy, audioFile);
/////// Transcribe the audio file. /////////////
// We want the object name in the cloud to be the original video file name with ".flac" extension.
string objectName = Path.GetFileNameWithoutExtension(videoFile) + ".flac";
TranscribeParameters transParams = new TranscribeParameters
{
audiofilePath = audioFile,
objectName = objectName,
GoogleCloudBucketName = config.GoogleCloudBucketName,
useAudioFileAlreadyInCloud = config.UseAudioFileAlreadyInCloud,
language = language,
MinSpeakerCount = 2,
MaxSpeakerCount = 6
// TODO Add "phrases" field: names of officers
};
Transcribed_Dto transcript = transcribeAudio.TranscribeAudioFile(transParams);
string stringValue = JsonConvert.SerializeObject(transcript, Formatting.Indented);
string outputJsonFile = Path.Combine(meetingFolder, "transcribed.json");
File.WriteAllText(outputJsonFile, stringValue);
}
public void TestMoveToCloudAndTranscribe(string language)
{
AudioProcessing audioProcessing = new AudioProcessing();
string baseName = "USA_ME_LincolnCounty_BoothbayHarbor_Selectmen_EN_2017-02-15";
string videoFile = Path.Combine(config.TestdataPath, baseName + ".mp4");
string outputFolder = Path.Combine(config.TestdataPath, "TestMoveToCloudAndTranscribe");
GMFileAccess.DeleteAndCreateDirectory(outputFolder);
string outputBasePath = Path.Combine(outputFolder, baseName);
string shortFile = outputBasePath + ".mp4";
string audioFile = outputBasePath + ".flac";
string jsonFile = outputBasePath + ".json";
// Extract short version
//SplitRecording splitRecording = new SplitRecording();
audioProcessing.ExtractPart(videoFile, shortFile, 60, 4 * 60);
// Extract audio.
audioProcessing.Extract(shortFile, audioFile);
// Transcribe
//TranscribeAudio ta = new TranscribeAudio(_config);
//TranscribeResultOrig response = new TranscribeResultOrig();
Transcribed_Dto response = new Transcribed_Dto();
// TODO - signature of TranscribeInCloud has changed.
// response = transcribe.MoveToCloudAndTranscribe(audioFile, baseName + ".flac", config.GoogleCloudBucketName, config.UseAudioFileAlreadyInCloud, language);
string stringValue = JsonConvert.SerializeObject(response, Formatting.Indented);
File.WriteAllText(outputBasePath + "-rsp.json", stringValue);
// Modify Transcript json format
//ModifyTranscriptJson_1 mt = new ModifyTranscriptJson_1();
ModifyTranscriptJson mt = new ModifyTranscriptJson();
//FixasrViewModel fixasr = mt.Modify(response);
EditMeeting_Dto meetingEditDto = mt.Modify(response);
// Create JSON file
//stringValue = JsonConvert.SerializeObject(fixasr, Formatting.Indented);
stringValue = JsonConvert.SerializeObject(meetingEditDto, Formatting.Indented);
File.WriteAllText(jsonFile, stringValue);
}
