public async Task StartTranscriptionAsync(Message message)
{
if (message == null)
{
throw new ArgumentNullException(nameof(message));
}
var busMessage = JsonConvert.DeserializeObject <ServiceBusMessage>(Encoding.UTF8.GetString(message.Body));
var audioFileName = StorageUtilities.GetFileNameFromUri(busMessage.Data.Url);
var startDateTime = DateTime.UtcNow;
// Check if language identification is required:
var secondaryLocale = StartTranscriptionEnvironmentVariables.SecondaryLocale;
if (!string.IsNullOrEmpty(secondaryLocale) &&
!secondaryLocale.Equals("None", StringComparison.OrdinalIgnoreCase) &&
!secondaryLocale.Equals(Locale, StringComparison.OrdinalIgnoreCase))
{
secondaryLocale = secondaryLocale.Split('|')[0].Trim();
Logger.LogInformation($"Primary locale: {Locale}");
Logger.LogInformation($"Secondary locale: {secondaryLocale}");
var languageID = new LanguageIdentification(StartTranscriptionEnvironmentVariables.AzureSpeechServicesKey, StartTranscriptionEnvironmentVariables.AzureSpeechServicesRegion);
var fileExtension = Path.GetExtension(audioFileName);
var sasUrl = await StorageUtilities.CreateSASAsync(StartTranscriptionEnvironmentVariables.AzureWebJobsStorage, busMessage.Data.Url, Logger).ConfigureAwait(false);
var byteArray = await StorageUtilities.DownloadFileFromSAS(sasUrl).ConfigureAwait(false);
var identifiedLocale = await languageID.DetectLanguage(byteArray, fileExtension, Locale, secondaryLocale).ConfigureAwait(false);
Logger.LogInformation($"Identified locale: {identifiedLocale}");
Locale = identifiedLocale;
}
await StartBatchTranscriptionJobAsync(new[] { message }, audioFileName, startDateTime).ConfigureAwait(false);
}
private async Task StartBatchTranscriptionJobAsync(IEnumerable <Message> messages, string jobName, DateTime startDateTime)
{
if (messages == null || !messages.Any())
{
Logger.LogError("Invalid service bus message(s).");
return;
}
var locationString = string.Empty;
var serviceBusMessages = messages.Select(message => JsonConvert.DeserializeObject <ServiceBusMessage>(Encoding.UTF8.GetString(message.Body)));
var modelIds = new List <Guid>();
try
{
var properties = GetTranscriptionPropertyBag();
modelIds = GetModelIds();
var audioFileUrls = new List <string>();
foreach (var serviceBusMessage in serviceBusMessages)
{
var audioFileUrl = await StorageUtilities.CreateSASAsync(StartTranscriptionEnvironmentVariables.AzureWebJobsStorage, serviceBusMessage.Data.Url, Logger).ConfigureAwait(false);
audioFileUrls.Add(audioFileUrl);
}
var client = new BatchClient(Subscription.SubscriptionKey, Subscription.LocationUri.AbsoluteUri, Logger);
var transcriptionLocation = await client.PostTranscriptionAsync(
jobName,
"StartByTimerTranscription",
Locale,
properties,
audioFileUrls,
modelIds).ConfigureAwait(false);
Logger.LogInformation($"Location: {transcriptionLocation}");
locationString = transcriptionLocation.ToString();
}
catch (WebException e)
{
if (BatchClient.IsThrottledOrTimeoutStatusCode(((HttpWebResponse)e.Response).StatusCode))
{
Logger.LogError($"Throttled or timeout while creating post, re-enqueueing transcription start. Message: {e.Message}");
var startTranscriptionSBConnectionString = StartTranscriptionEnvironmentVariables.StartTranscriptionServiceBusConnectionString;
foreach (var message in messages)
{
ServiceBusUtilities.SendServiceBusMessageAsync(startTranscriptionSBConnectionString, message, Logger, 2).GetAwaiter().GetResult();
}
return;
}
else
{
Logger.LogError($"Failed with Webexception. Write message for job with name {jobName} to report file.");
var errorTxtName = jobName + ".txt";
var exceptionMessage = e.Message;
using (var reader = new StreamReader(e.Response.GetResponseStream()))
{
var responseMessage = await reader.ReadToEndAsync().ConfigureAwait(false);
exceptionMessage += "\n" + responseMessage;
}
await WriteFailedJobLogToStorageAsync(serviceBusMessages, exceptionMessage, jobName).ConfigureAwait(false);
return;
}
}
catch (TimeoutException e)
{
Logger.LogError($"Timeout while creating post, re-enqueueing transcription start. Message: {e.Message}");
var startTranscriptionSBConnectionString = StartTranscriptionEnvironmentVariables.StartTranscriptionServiceBusConnectionString;
foreach (var message in messages)
{
ServiceBusUtilities.SendServiceBusMessageAsync(startTranscriptionSBConnectionString, message, Logger, 2).GetAwaiter().GetResult();
}
return;
}
catch (Exception e)
{
Logger.LogError($"Failed with Exception {e} and message {e.Message}. Write message for job with name {jobName} to report file.");
await WriteFailedJobLogToStorageAsync(serviceBusMessages, e.Message, jobName).ConfigureAwait(false);
return;
}
var reenqueueingTimeInSeconds = MapDatasetSizeToReenqueueingTimeInSeconds(StartTranscriptionEnvironmentVariables.AudioDatasetSize);
var transcriptionMessage = new PostTranscriptionServiceBusMessage(
Subscription,
locationString,
jobName,
startDateTime.ToString(CultureInfo.InvariantCulture),
Locale,
modelIds.Any(),
StartTranscriptionEnvironmentVariables.AddSentimentAnalysis,
StartTranscriptionEnvironmentVariables.AddEntityRedaction,
0,
reenqueueingTimeInSeconds,
reenqueueingTimeInSeconds);
var fetchTranscriptionSBConnectionString = StartTranscriptionEnvironmentVariables.FetchTranscriptionServiceBusConnectionString;
Logger.LogInformation($"FetchTranscriptionServiceBusConnectionString from settings: {fetchTranscriptionSBConnectionString}");
try
{
ServiceBusUtilities.SendServiceBusMessageAsync(fetchTranscriptionSBConnectionString, transcriptionMessage.CreateMessageString(), Logger, reenqueueingTimeInSeconds).GetAwaiter().GetResult();
}
catch (Exception e)
{
Logger.LogError($"Failed with Exception {e} and message {e.Message}. Write message for job with name {jobName} to report file.");
await WriteFailedJobLogToStorageAsync(serviceBusMessages, e.Message, jobName).ConfigureAwait(false);
}
Logger.LogInformation($"Fetch transcription queue informed about job at: {jobName}");
}