/// <summary>
/// Convert text to speech to an audio file then save it to a S3 bucket
/// </summary>
/// <param name="request"></param>
/// <returns></returns>
public async Task <ConvertTextResponse> AddTextToSpeechFileToBucket(ConvertTextRequest request)
{
if (string.IsNullOrEmpty(request.Content))
{
throw new ArgumentNullException("Content cannot be empty");
}
_logger.Info($"Content requested to convert: {request.Content}");
var pollyRequest = new SynthesizeSpeechRequest {
OutputFormat = OutputFormat.Mp3,
Text = request.Content,
TextType = "text",
VoiceId = VoiceId.Amy, // https://docs.aws.amazon.com/sdkfornet/v3/apidocs/items/Polly/TVoiceId.html
LanguageCode = LanguageCode.EnUS // https://docs.aws.amazon.com/sdkfornet/v3/apidocs/items/Polly/TLanguageCode.html
};
// Send request to Amazon Polly to convert text to audio.
var synthesizeSpeechResponse = await _pollyClient.SynthesizeSpeechAsync(pollyRequest);
if (synthesizeSpeechResponse == null ||
synthesizeSpeechResponse.HttpStatusCode != (HttpStatusCode)200)
{
throw new Exception("Text could not be converted to audio.");
}
_logger.Info($"Requested characters was: {synthesizeSpeechResponse.RequestCharacters}");
// Stream the audio from Amazon Polly to a S3 bucket using the filename provided by the request
PutObjectResponse s3Response;
using (var memoryStream = new MemoryStream()) {
synthesizeSpeechResponse.AudioStream.CopyTo(memoryStream);
s3Response = await _s3Client.PutObjectAsync(new PutObjectRequest {
BucketName = _s3BucketName,
Key = request.FileName,
InputStream = memoryStream
});
}
if (s3Response == null ||
s3Response.HttpStatusCode != (HttpStatusCode)200)
{
throw new Exception("Unable to save audio file to s3");
}
// Publish to a SNS topic that a new audio file has been saved.
var publishRequest = new PublishRequest {
TopicArn = _notificationSnsTopic,
Message = "An audio file was saved",
Subject = "New MP3 File"
};
await _snsClient.PublishAsync(publishRequest);
return(new ConvertTextResponse {
FileName = request.FileName
});
}
/// <summary>
/// Calls the Amazon Polly SynthesizeSpeechAsync method to convert text
/// to speech.
/// </summary>
/// <param name="client">The Amazon Polly client object used to connect
/// to the Amazon Polly service.</param>
/// <param name="text">The text to convert to speech.</param>
/// <returns>A SynthesizeSpeechResponse object that includes an AudioStream
/// object with the converted text.</returns>
private static async Task <SynthesizeSpeechResponse> PollySynthesizeSpeech(IAmazonPolly client, string text)
{
var synthesizeSpeechRequest = new SynthesizeSpeechRequest()
{
OutputFormat = OutputFormat.Mp3,
VoiceId = VoiceId.Joanna,
Text = text,
};
var synthesizeSpeechResponse =
await client.SynthesizeSpeechAsync(synthesizeSpeechRequest);
return(synthesizeSpeechResponse);
}
public async static Task <string> PollyDemo(IAmazonPolly pollyClient, IAmazonS3 S3Client, string text)
{
string result = null;
SynthesizeSpeechRequest synthesizeRequest = new SynthesizeSpeechRequest()
{
LanguageCode = LanguageCode.EnUS,
OutputFormat = "mp3",
SampleRate = "8000",
Text = text,
TextType = "text",
VoiceId = "Joanna"
};
try
{
Task <SynthesizeSpeechResponse> synthesizeTask = pollyClient.SynthesizeSpeechAsync(synthesizeRequest);
SynthesizeSpeechResponse syntheizeResponse = await synthesizeTask;
Console.WriteLine(syntheizeResponse.ContentType);
Console.WriteLine(syntheizeResponse.RequestCharacters);
using (MemoryStream ms = new MemoryStream())
{
syntheizeResponse.AudioStream.CopyTo(ms);
Console.WriteLine(ms.Length);
// Upload image to S3 bucket
string bucketName = "reinvent-indiamazones";
//string key = dto.text;
string key = "pollytest";
await Task.Run(() => S3Util.UploadToS3(S3Client, bucketName, key, ms));
// TODO : need to check the file exists in S3
result = S3Util.GetPresignedURL(S3Client, bucketName, key);
}
//syntheizeResponse.AudioStream.CopyTo(result);
//result.Flush();
}
catch (AmazonPollyException pollyException)
{
Console.WriteLine(pollyException.Message, pollyException.InnerException);
}
return(result);
}
private Amazon.Polly.Model.SynthesizeSpeechResponse CallAWSServiceOperation(IAmazonPolly client, Amazon.Polly.Model.SynthesizeSpeechRequest request)
{
Utils.Common.WriteVerboseEndpointMessage(this, client.Config, "Amazon Polly", "SynthesizeSpeech");
try
{
#if DESKTOP
return(client.SynthesizeSpeech(request));
#elif CORECLR
return(client.SynthesizeSpeechAsync(request).GetAwaiter().GetResult());
#else
#error "Unknown build edition"
#endif
}
catch (AmazonServiceException exc)
{
var webException = exc.InnerException as System.Net.WebException;
if (webException != null)
{
throw new Exception(Utils.Common.FormatNameResolutionFailureMessage(client.Config, webException.Message), webException);
}
throw;
}
}
/// <summary>
///
/// </summary>
/// <param name="input"></param>
/// <param name="context"></param>
/// <returns></returns>
public async Task FunctionHandler(SNSEvent @event, ILambdaContext context)
{
var logger = context.Logger;
var postId = @event.Records[0].Sns.Message;
logger.LogLine($"Text to Speech function. Post ID in DynamoDB: {postId}");
// Retrieving information about the post from DynamoDB table
var tableName = Environment.GetEnvironmentVariable("DB_TABLE_NAME");
var queryResult = await _dbClient.QueryAsync(new QueryRequest
{
TableName = tableName,
KeyConditions = new Dictionary <string, Condition>
{
{
"id", new Condition()
{
ComparisonOperator = ComparisonOperator.EQ,
AttributeValueList = new List <AttributeValue>
{
new AttributeValue(postId)
}
}
}
}
});
var post = queryResult.Items[0];
var text = post["text"].S;
var voice = post["voice"].S;
// Chunk and Synthesize text into speech.
var chunkSize = 1000;
var textBlocks = text.Chunk(chunkSize).ToArray();
var path = System.IO.Path.Combine("/tmp/", postId);
for (var i = 0; i < textBlocks.Length; i++)
{
var textBlock = textBlocks[i];
var pollyResponse = await _pollyClient.SynthesizeSpeechAsync(new SynthesizeSpeechRequest
{
OutputFormat = "mp3",
Text = textBlock,
VoiceId = voice
});
if (pollyResponse?.AudioStream != null)
{
using (var fileStream = System.IO.File.Open(path, System.IO.FileMode.Append))
{
await pollyResponse.AudioStream.CopyToAsync(fileStream);
}
logger.LogLine($"Finished synthesizing text chunck {i + 1}");
}
}
//Put synthesized audio into s3 bucket
var bucketName = Environment.GetEnvironmentVariable("BUCKET_NAME");
await _s3Client.UploadObjectFromFilePathAsync(bucketName, $"{postId}.mp3", path, null);
await _s3Client.PutACLAsync(new PutACLRequest
{
BucketName = bucketName,
Key = $"{postId}.mp3",
CannedACL = S3CannedACL.PublicRead
});
var bucketLocation = await _s3Client.GetBucketLocationAsync(new GetBucketLocationRequest
{
BucketName = bucketName
});
var region = bucketLocation.Location;
// Update DynamoDB with the audio file url and the "UPDATED" status
var urlBeginning = region == null || string.IsNullOrWhiteSpace(region.Value)
? "https://s3.amazonaws.com/"
: $"https://s3-{region.Value}.amazonaws.com/";
var url = string.Concat(urlBeginning, bucketName, "/", postId, ".mp3");
await _dbClient.UpdateItemAsync(new UpdateItemRequest
{
TableName = tableName,
Key = new Dictionary <string, AttributeValue>
{
{ "id", new AttributeValue(postId) }
},
UpdateExpression = "SET #statusAtt = :statusValue, #urlAtt = :urlValue",
ExpressionAttributeValues = new Dictionary <string, AttributeValue>
{
{ ":statusValue", new AttributeValue("UPDATED") },
{ ":urlValue", new AttributeValue(url) }
},
ExpressionAttributeNames = new Dictionary <string, string>
{
{ "#statusAtt", "status" },
{ "#urlAtt", "url" }
}
});
logger.LogLine($"Finished updating post ,ID: {postId}, in table {tableName}");
}
public AudioInfo Synthesize(string text)
{
Dictionary <uint, byte[]> audioBytes = new Dictionary <uint, byte[]>();
var ulawFormat = WaveFormat.CreateMuLawFormat(8000, 1);
string tempFile = Path.GetTempFileName();
try
{
SynthesizeSpeechRequest sreq = new SynthesizeSpeechRequest();
sreq.Text = text;
sreq.OutputFormat = OutputFormat.Mp3;
sreq.VoiceId = VoiceId.Salli;
SynthesizeSpeechResponse sres = _amazonPolly.SynthesizeSpeechAsync(sreq).GetAwaiter().GetResult();
using (var pollyMemoryStream = new MemoryStream())
{
sres.AudioStream.CopyTo(pollyMemoryStream);
pollyMemoryStream.Flush();
pollyMemoryStream.Position = 0;
using (Mp3FileReader reader = new Mp3FileReader(pollyMemoryStream, wave => new DmoMp3FrameDecompressor(wave)))
{
using (WaveStream pcmStream = WaveFormatConversionStream.CreatePcmStream(reader))
{
WaveFileWriter.CreateWaveFile(tempFile, pcmStream);
}
}
}
var pcmFormat = new WaveFormat(8000, 16, 1);
List <byte[]> allBytes = new List <byte[]>();
using (WaveFormatConversionStream pcmStm = new WaveFormatConversionStream(pcmFormat, new WaveFileReader(tempFile)))
{
using (WaveFormatConversionStream ulawStm = new WaveFormatConversionStream(ulawFormat, pcmStm))
{
byte[] buffer = new byte[160];
int bytesRead = ulawStm.Read(buffer, 0, 160);
while (bytesRead > 0)
{
byte[] sample = new byte[bytesRead];
Array.Copy(buffer, sample, bytesRead);
allBytes.Add(sample);
bytesRead = ulawStm.Read(buffer, 0, 160);
}
int secondsToAdd = 1;
var silentBytes = new byte[ulawStm.WaveFormat.AverageBytesPerSecond * secondsToAdd];
allBytes.Add(silentBytes);
}
}
AudioInfo ai = new AudioInfo()
{
AudioData = allBytes.SelectMany(p => p).ToArray(),
Codec = Codec.G711U
};
return(ai);
//string fileName = @".\TestAudio\output-from-polly-mp3-then-wav.wav";
//using (WaveFileWriter writer = new WaveFileWriter(fileName, ulawFormat))
//{
// var testSequence = audioBytes.SelectMany(p => p.Value).ToArray();
// writer.Write(testSequence, 0, testSequence.Length);
//}
}
finally
{
try
{
File.Delete(tempFile);
}
catch { }
}
}
/// <summary>
/// A function handling the APIGatewayProxyRequest
/// </summary>
/// <param name="request"></param>
/// <param name="context"></param>
/// <returns></returns>
public async Task <APIGatewayProxyResponse> FunctionHandler(APIGatewayProxyRequest request, ILambdaContext context)
{
var synthesizeSpeechResponse = new SynthesizeSpeechResponse();
try
{
synthesizeSpeechResponse = await _pollyClient.SynthesizeSpeechAsync(new SynthesizeSpeechRequest()
{
Engine = Engine.Neural,
LanguageCode = LanguageCode.EnUS,
OutputFormat = OutputFormat.Mp3,
SampleRate = "24000",
Text = "Hi, my name is Szilard. I am from Brasov and I work at Siemens.",
TextType = TextType.Text,
VoiceId = VoiceId.Joanna
});
}
catch (AmazonPollyException ex)
{
context.Logger.LogLine(ex.Message);
return(new APIGatewayProxyResponse()
{
StatusCode = 500,
Body = ex.Message,
Headers = new Dictionary <string, string>()
{
{ "Content-Type", "text/plain" },
{ "Access-Control-Allow-Origin", ALLOWED_ORIGIN }
}
});
}
//
// put audio stream into S3
//
using (MemoryStream stream = new MemoryStream())
{
try
{
await synthesizeSpeechResponse.AudioStream.CopyToAsync(stream);
}
catch (Exception ex)
{
context.Logger.LogLine(ex.Message);
return(new APIGatewayProxyResponse()
{
StatusCode = 500,
Body = ex.Message,
Headers = new Dictionary <string, string>()
{
{ "Content-Type", "text/plain" },
{ "Access-Control-Allow-Origin", ALLOWED_ORIGIN }
}
});
}
try
{
await _s3Client.PutObjectAsync(new PutObjectRequest()
{
InputStream = stream,
BucketName = TARGET_BUCKET,
Key = Path.Combine("audios", DateTime.Now.ToString("dd-MM-yyyy_HH-mm-ss-ffff") + ".mp3")
});
}
catch (AmazonS3Exception ex)
{
context.Logger.LogLine(ex.Message);
return(new APIGatewayProxyResponse()
{
StatusCode = 500,
Body = ex.Message,
Headers = new Dictionary <string, string>()
{
{ "Content-Type", "text/plain" },
{ "Access-Control-Allow-Origin", ALLOWED_ORIGIN }
}
});
}
}
//
// return successfull response
//
return(new APIGatewayProxyResponse()
{
StatusCode = 200,
Headers = new Dictionary <string, string>()
{
{ "Access-Control-Allow-Origin", ALLOWED_ORIGIN }
}
});
}
