private string CreateTextPartFilesFromSnippet(string ExportFolder, TrackSnippetViewModel snip, MessageBoxResult yesIfVoice, ref int index)
{
var transcriptionsFileText = "";
foreach (var tp in snip.TextParts)
{
if (!tp.IsOK || string.IsNullOrWhiteSpace(tp.Text))
{
continue;
}
var fileNumberString = $"{index.ToString().PadLeft(4, '0')}";
var fileName = Path.Combine(ExportFolder, $"{fileNumberString}.wav"); //Path.GetFileName(snip.FilePath);
if (yesIfVoice == MessageBoxResult.Yes)
{
transcriptionsFileText += $"{fileNumberString}\t{tp.Text.Trim()}{Environment.NewLine}"; // is for Custom Voice
}
else
{
transcriptionsFileText += $"{fileNumberString}.wav\t{tp.Text.Trim()}{Environment.NewLine}"; // is for Acoutic Dataset
}
WavFileUtils.TakeClipAddSilence(snip.FilePath, TimeSpan.FromMilliseconds(800), TimeSpan.FromMilliseconds(tp.StartMills), TimeSpan.FromMilliseconds(tp.TextWidth), fileName);
index++;
}
return(transcriptionsFileText);
}
private void MakeSnippetAudioFiles(
TrackSnippetViewModel snippet,
int ix)
{
var newProjectFolder = Path.Combine(Settings.Default.ProjectsFolder, SelectedProject.Name, $"{DateTime.Now.ToString("ddMMyyyyHHmmss")}");
Directory.CreateDirectory(newProjectFolder);
snippet.FilePath = Path.Combine(newProjectFolder, $"{Path.GetFileNameWithoutExtension(AudioFilePath)}_{ix++}.wav");
var startTime = TimeSpan.FromTicks(snippet.AudioSnippet.OffsetInTicks);
WavFileUtils.TakeClipAddSilence(AudioFilePath, TimeSpan.Zero, startTime, snippet.AudioSnippet.Duration, snippet.FilePath);
}
private async Task GetTranscriptionFromAzure(TaskCompletionSource <int> stopRecognition, List <TrackSnippetViewModel> newCollection)
{
// Creates an instance of a speech config with specified subscription key and service region.
// Replace with your own subscription key and service region (e.g., "westus").
var config = SpeechConfig.FromSubscription(TranscribeEndpointsVM.SelectedTranscribeEndpoint.Key, TranscribeEndpointsVM.SelectedTranscribeEndpoint.Region);
if (!string.IsNullOrWhiteSpace(TranscribeEndpointsVM.SelectedTranscribeEndpoint.Endpoint))
{
config.EndpointId = TranscribeEndpointsVM.SelectedTranscribeEndpoint.Endpoint;
}
config.OutputFormat = OutputFormat.Detailed;
var log = "";
double OnePercent = 100D / CurrentAudioVM.Duration.TimeSpan.Ticks;
var tLog = new transLog {
Started = DateTime.Now, FilePath = AudioFilePath
};
try
{
if (Path.GetExtension(AudioFilePath).ToLower() != ".wav")
{
// extract the WAV
var wavFileName = Path.Combine(Path.GetDirectoryName(AudioFilePath), Path.GetFileNameWithoutExtension(AudioFilePath) + ".wav");
WavFileUtils.ExtractWavFromMedia(AudioFilePath, wavFileName);
AudioFilePath = wavFileName;
}
using (var audioInput = AudioConfig.FromWavFileInput(AudioFilePath))
{
using (var recognizer = new SpeechRecognizer(config, audioInput))
{
// Subscribes to events.
recognizer.Recognizing += (s, e) =>
{
transLogRow dyn = new transLogRow();
var txtCnt = textPartsAll.Count;
var dur = e.Result.Duration;
var totDone = TimeSpan.FromTicks(e.Result.OffsetInTicks + dur.Ticks);
var perc = OnePercent * (dur.Ticks + e.Result.OffsetInTicks);
TranscribeInfo = $"{Math.Floor(perc)}%: {newCollection.Count + 1}: {totDone.Hours}:{totDone.Minutes}:{totDone.Seconds}: {txtCnt}: {e.Result.Text}";
dyn.Args = e;
var textBit = e.Result.Text;
var ary = textBit.Split(' ');
var lastGoodIx = 0;
var goodTextParts = new List <TextPart>();
var isChange = false;
for (var a = 0; a < ary.Length; a++)
{
if (a > txtCnt - 1)
{
dyn.Notes += $"[new={txtCnt - a}]";
break;
}
if (ary[a].Trim() != textPartsAll[a].Text.Trim())
{
dyn.Notes += $"[chg={a}:{textPartsAll[a].Text}={ary[a]}]";
textPartsAll[a].Text = ary[a];
isChange = true;
}
goodTextParts.Add(textPartsAll[a]);
lastGoodIx = a;
}
if (lastGoodIx < textPartsAll.Count - 1)
{
dyn.Notes += $"[lth chg]";
}
double detectedEstimate;
try
{
if (goodTextParts.Count >= ary.Length)
{
return; // nothing new
}
var newBit = "";
for (var x = goodTextParts.Count; x < ary.Length; x++)
{
newBit += ary[x] + " ";
}
// 200 magic number - identification time about 200 mills
detectedEstimate = e.Result.Duration.TotalMilliseconds - 200;
var sinceLastEvent = detectedEstimate - lastMillisecondsTotal;
if (sinceLastEvent < 1)
{
sinceLastEvent = 1;
}
var cnt = goodTextParts.Count;
if (cnt > 1 && !isChange)
{
if (sinceLastEvent < ADJUSTED_SIZE_MAX_WORD_LENGTH)
{
dyn.Notes += $"[adj={cnt - 1}, snc={sinceLastEvent}, lgt={goodTextParts[cnt - 1].TextWidth} - {ADJUSTED_SIZE_MAX_WORD_LENGTH - sinceLastEvent}]";
goodTextParts[cnt - 1].TextWidth -= (ADJUSTED_SIZE_MAX_WORD_LENGTH - sinceLastEvent);
}
}
// Finally, save any new bit(s)
goodTextParts.Add(new TextPart {
Text = newBit.Trim(), StartMills = detectedEstimate, TextWidth = ADJUSTED_SIZE_MAX_WORD_LENGTH
});
}
catch (Exception ex)
{
throw;
}
finally
{
log += JsonConvert.SerializeObject(dyn) + Environment.NewLine;
}
lastMillisecondsTotal = detectedEstimate; // start point for next text
// lastCharLength = textBit.Length;
textPartsAll = goodTextParts;
};
recognizer.Recognized += (s, e) =>
{
if (e.Result.Reason == ResultReason.RecognizedSpeech)
{
//Console.WriteLine($"RECOGNIZED: Text={e.Result.Text}");
var snippet = new TrackSnippetViewModel()
{
AudioSnippet = e.Result,
Scale = scale,
Gain = gain,
RawText = e.Result.Text,
OffsetInTicks = e.Result.OffsetInTicks
};
snippet.DurationMilliseconds = e.Result.Duration.TotalMilliseconds; //WavFileUtils.GetSoundLength(AudioFilePath);
//var rawAry = e.Result.Text.Split(' ');
//if (rawAry.Length == textPartsAll.Count)
//{
// for(var a = 0; a < rawAry.Length; a++)
// {
// textPartsAll[a].Text = rawAry[a];
// }
//}
//else
//{
// // to do, merge tidied text back into recognised text
//}
snippet.TextParts = new ObservableCollection <TextPart>(textPartsAll);
parentControl.Dispatcher.Invoke(() => newCollection.Add(snippet));
textPartsAll = new List <TextPart>();
lastMillisecondsTotal = 0;
}
else if (e.Result.Reason == ResultReason.NoMatch)
{
Console.WriteLine($"NOMATCH: Speech could not be recognized.");
}
};
recognizer.Canceled += (s, e) =>
{
Console.WriteLine($"CANCELED: Reason={e.Reason}");
if (e.Reason == CancellationReason.Error)
{
MessageBox.Show($"FAILED: Are you using a valid subscription key and region? ErrorCode = {e.ErrorCode}: ErrorDetails = {e.ErrorDetails}");
}
stopRecognition.TrySetResult(0);
};
recognizer.SessionStarted += (s, e) => Console.WriteLine("\n Session started event.");
recognizer.SessionStopped += (s, e) =>
{
Console.WriteLine("\n Session stopped event.");
Console.WriteLine("\nStop recognition.");
stopRecognition.TrySetResult(0);
};
// Starts continuous recognition. Uses StopContinuousRecognitionAsync() to stop recognition.
await recognizer.StartContinuousRecognitionAsync().ConfigureAwait(false);
// Waits for completion.
// Use Task.WaitAny to keep the task rooted.
Task.WaitAny(new[] { stopRecognition.Task });
// Stops recognition.
await recognizer.StopContinuousRecognitionAsync().ConfigureAwait(false);
}
}
}
catch (Exception excp)
{
Debug.WriteLine($"{excp}");
}
finally
{
if (Settings.Default.TranscriptionLogging)
{
try
{
var logFile = $"TranscriptionLog_{DateTime.Now.ToString($"{DateTime.Now.ToString("ddMMyyyyHHmmss")}")}.txt";
logFile = Path.Combine(Path.GetFullPath(SettingsVM.ProjectsFolder), logFile);
File.WriteAllText(logFile, log);
tLog.LogPath = logFile;
SelectedProject.TranscribeLogs.Add(tLog);
Debug.WriteLine($"Log file created at {logFile}");
}
catch (Exception ex)
{
throw;
}
}
}
}