public async Task <ConvertResponse> ExtractTextComputerVision(ConvertResponse response)
{
try
{
var key = _config["computerVision:key"];
var endpoint = _config["computerVision:endpoint"];
ComputerVisionClient computerVision = new ComputerVisionClient(new ApiKeyServiceClientCredentials(key))
{
Endpoint = endpoint
};
var analysis = await computerVision.RecognizePrintedTextAsync(true, response.Request.UploadBlobUrl + _config["storage:sas"]);
var text = new StringBuilder();
foreach (var region in analysis.Regions)
{
foreach (var line in region.Lines)
{
foreach (var word in line.Words)
{
text.Append(word.Text + " ");
}
text.AppendLine();
}
}
var transcriptBlobName = Path.GetFileNameWithoutExtension(response.Request.BlobName) + ".txt";
var blobClient = new BlobStorageClient(_config);
var textBlob = blobClient.GetBlobBlock("transcripts", transcriptBlobName);
response.TranscriptBlobUrl = textBlob.Uri.AbsoluteUri;
response.Transcript = text.ToString().Trim();
await textBlob.UploadTextAsync(response.Transcript);
}
catch (Exception ex)
{
Console.WriteLine(ex.ToString());
Console.WriteLine(ex.Message);
response.ErrorMessage = ex.ToString();
}
return(response);
}
public async Task <ConvertResponse> ExtractTextBatchRead(ConvertResponse response)
{
try
{
var key = _config["computerVision:key"];
var endpoint = _config["computerVision:endpoint"];
ComputerVisionClient computerVision = new ComputerVisionClient(new ApiKeyServiceClientCredentials(key))
{
Endpoint = endpoint
};
var operationInfo = await computerVision.BatchReadFileAsync(response.Request.UploadBlobUrl + _config["storage:sas"]);
var result = new ReadOperationResult();
var operationId = operationInfo.OperationLocation.Split('/').Last();
while (result.Status != TextOperationStatusCodes.Failed && result.Status != TextOperationStatusCodes.Succeeded)
{
await Task.Delay(500);
result = await computerVision.GetReadOperationResultAsync(operationId);
}
if (result.Status == TextOperationStatusCodes.Failed)
{
response.ErrorMessage = $"Text translation failed.";
return(response);
}
var text = new StringBuilder();
foreach (var page in result.RecognitionResults)
{
Line lastLine = null;
foreach (var line in page.Lines)
{
// if (lastLine?.Words.Count >= 4)
// {
// text.Append($" {line.Text}");
// }
// else
// {
text.Append(Environment.NewLine + line.Text);
// }
lastLine = line;
}
}
Console.WriteLine();
Console.WriteLine(text.ToString());
var transcriptBlobName = Path.GetFileNameWithoutExtension(response.Request.BlobName) + ".txt";
var blobClient = new BlobStorageClient(_config);
var textBlob = blobClient.GetBlobBlock("transcripts", transcriptBlobName);
response.TranscriptBlobUrl = textBlob.Uri.AbsoluteUri;
response.Transcript = text.ToString().Trim();
await textBlob.UploadTextAsync(response.Transcript);
}
catch (Exception ex)
{
Console.WriteLine(ex.ToString());
Console.WriteLine(ex.Message);
response.ErrorMessage = ex.ToString();
}
return(response);
}
public async Task <ConvertResponse> ConvertTextToSpeechSsml(ConvertResponse response)
{
var ext = new FileInfo(response.Request.BlobName).Extension;
var waveName = response.Request.BlobName.Replace(ext, ".wav");
try
{
string accessToken = null;
using (var client = new HttpClient())
{
client.DefaultRequestHeaders.Add("Ocp-Apim-Subscription-Key", _config["speech:key"]);
var tokenFetchUri = $"https://{_config["speech:region"]}.api.cognitive.microsoft.com/sts/v1.0/issueToken";
var uriBuilder = new UriBuilder(tokenFetchUri);
var result = await client.PostAsync(uriBuilder.Uri.AbsoluteUri, null).ConfigureAwait(false);
accessToken = await result.Content.ReadAsStringAsync().ConfigureAwait(false);
}
var r = response.Request;
var host = $"https://{_config["speech:region"]}.tts.speech.microsoft.com/cognitiveservices/v1";
var prosody = $"<prosody rate='{r.Speed}.00%' pitch='{r.Pitch}' volume='{r.Volume}'>";
response.Transcript = response.Transcript.Replace("&", "and");
var maxChars = 990;
var transcript = response.Transcript;
if (transcript.Length > 990)
{
var intro = "This transcript is too long and has been shortened automatically." + Environment.NewLine;
transcript = intro + transcript.Substring(0, maxChars - intro.Length);
}
string body = $@"<speak version='1.0' xmlns='https://www.w3.org/2001/10/synthesis' xml:lang='en-US'>
<voice name='Microsoft Server Speech Text to Speech Voice (en-US, {r.VoiceToUse})'>{prosody}" +
transcript + "</prosody></voice></speak>";
Console.WriteLine();
Console.WriteLine(body);
using (var client = new HttpClient())
{
using (var request = new HttpRequestMessage())
{
request.Method = HttpMethod.Post;
request.RequestUri = new Uri(host);
request.Content = new StringContent(body, Encoding.UTF8, "application/ssml+xml");
request.Headers.Add("Authorization", "Bearer " + accessToken);
request.Headers.Add("Connection", "Keep-Alive");
request.Headers.Add("User-Agent", _config["speech:resourceName"]);
request.Headers.Add("X-Microsoft-OutputFormat", "riff-24khz-16bit-mono-pcm");
Console.WriteLine("Calling the TTS service. Please wait... \n");
using (var speechResponse = await client.SendAsync(request).ConfigureAwait(false))
{
speechResponse.EnsureSuccessStatusCode();
using (var dataStream = await speechResponse.Content.ReadAsStreamAsync().ConfigureAwait(false))
{
var fileName = Path.GetTempPath() + "/temp.wav";
Console.WriteLine("Your speech file is being written to temp file...");
using (var fileStream = new FileStream(fileName, FileMode.Create, FileAccess.Write, FileShare.Write))
{
await dataStream.CopyToAsync(fileStream).ConfigureAwait(false);
fileStream.Close();
}
var blobClient = new BlobStorageClient(_config);
var audioBlob = blobClient.GetBlobBlock("speechresults", waveName);
await audioBlob.UploadFromFileAsync(fileName);
File.Delete(fileName);
response.SpeechAudioBlobUrl = audioBlob.Uri.AbsoluteUri;
response.IsSuccessful = true;
}
}
}
}
}
catch (Exception ex)
{
Console.WriteLine(ex.ToString());
Console.WriteLine(ex.Message);
response.ErrorMessage = ex.ToString();
}
return(response);
}
