saherelm/xSaherElmAiApi
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Complete Audi Transcript using NAudio and Whisper ...

Browse Source
This commit is contained in:
saherelm
2026-10-04 11:42:27 +03:30
parent 70c2d5a649
commit e6e5369b95
5 changed files with 225 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Providers/Extractors/XAudioFileContentExtractor.cs
+196 -21
View File
@@ -2,11 +2,15 @@ using System;
using System.IO;
using System.Linq;
using Whisper.net;
using NAudio.Wave;
using System.Threading;
using xAiModels.Models;
using xCommons.Extensions;
using xAiApi.Configurations;
using xExceptions.Constants;
using System.Threading.Tasks;
using NAudio.MediaFoundation;
using xAiApi.Interfaces.Extractors;
using DocumentFormat.OpenXml.EMMA;
namespace xAiApi.Providers.Extractors
{
@@ -24,29 +28,30 @@ namespace xAiApi.Providers.Extractors
"audio/mp3",
"audio/wav",
"audio/wave",
"audio/x-wav",
"audio/ogg",
"audio/m4a",
"audio/mp4",
"audio/aac",
"audio/x-m4a",
"audio/webm"
];
private readonly string language;
private readonly string whisperModelPath;
public XAudioFileContentExtractor() : this(
language: "fa",
whisperModelPath: "Models/ggml-base.bin"
)
{ }
private readonly XAiApiConfiguration configuration;
public XAudioFileContentExtractor(
string whisperModelPath = "Models/ggml-base.bin",
string language = "fa"
XAiApiConfiguration configuration
)
{
//
this.language = language;
this.whisperModelPath = whisperModelPath;
if (configuration.Transcript.IsNullOrDefault() ||
configuration.Transcript.ModelsPath.IsNullOrEmpty())
{
XException.InvalidConfiguration.Throw();
}
//
this.configuration = configuration;
}
/// <summary>
@@ -113,14 +118,15 @@ namespace xAiApi.Providers.Extractors
//
using var wavStream = await ConvertToWavAsync(
memoryStream,
cancellationToken
mimeType: mimeType,
inputStream: memoryStream,
cancellationToken: cancellationToken
);
//
using var factory = WhisperFactory.FromPath(whisperModelPath);
using var factory = WhisperFactory.FromPath(configuration.Transcript.ModelsPath);
using var processor = factory.CreateBuilder()
.WithLanguage(language)
.WithLanguage(configuration.Transcript.Language)
.Build();
//
@@ -143,17 +149,186 @@ namespace xAiApi.Providers.Extractors
return result;
}
//
#region Private ...
/// <summary>
/// Audio Format Converting ...
/// Audio Format Converting to WAV 16kHz Mono 16-bit PCM ...
/// </summary>
private async Task<Stream> ConvertToWavAsync(
Stream inputStream,
CancellationToken cancellationToken
string mimeType,
CancellationToken cancellationToken = default
)
{
//
// TODO: Complete this ...
return await Task.FromResult(inputStream);
var result = await Task.Run(() =>
{
//
var memoryStream = new MemoryStream();
inputStream.CopyTo(memoryStream);
memoryStream.Position = 0;
//
var targetFormat = new WaveFormat(16000, 16, 1);
string tempFilePath = null;
//
try
{
//
WaveStream reader = GetAudioReader(memoryStream, mimeType, ref tempFilePath);
if (reader == null)
{
//
XException.NotAllowed.Throw(
$"Unsupported audio format: {mimeType}"
);
}
//
var needsConversion =
reader.WaveFormat.SampleRate != 16000 ||
reader.WaveFormat.Channels != 1 ||
reader.WaveFormat.BitsPerSample != 16 ||
reader.WaveFormat.Encoding != WaveFormatEncoding.Pcm;
var outputStream = new MemoryStream();
if (!needsConversion)
{
//
reader.CopyTo(outputStream);
reader.Dispose();
}
else
{
//
MediaFoundationApi.Startup();
using var resampler = new MediaFoundationResampler(
reader,
targetFormat
);
resampler.ResamplerQuality = 60;
//
WaveFileWriter.WriteWavFileToStream(outputStream, resampler);
reader.Dispose();
}
//
outputStream.Position = 0;
return outputStream;
}
finally
{
//
if (!string.IsNullOrEmpty(tempFilePath) &&
File.Exists(tempFilePath))
{
//
try
{
File.Delete(tempFilePath);
}
catch { }
}
//
memoryStream.Dispose();
}
}, cancellationToken);
//
return result;
}
/// <summary>
/// Get appropriate WaveStream reader based on MIME type ...
/// </summary>
/// <param name="stream">Input stream (MemoryStream)</param>
/// <param name="mimeType">MIME type of audio file</param>
/// <param name="tempFilePath">Path to temp file (if created)</param>
/// <returns>WaveStream reader or null if unsupported</returns>
private WaveStream GetAudioReader(
MemoryStream stream,
string mimeType,
ref string tempFilePath
)
{
//
try
{
//
var normalizedMime = mimeType?.ToLowerInvariant() ?? string.Empty;
switch (normalizedMime)
{
//
case "audio/wav":
case "audio/wave":
case "audio/x-wav":
return new WaveFileReader(stream);
//
case "audio/mpeg":
case "audio/mp3":
return new Mp3FileReader(stream);
//
case "audio/m4a":
case "audio/mp4":
case "audio/aac":
case "audio/ogg":
case "audio/webm":
case "audio/x-m4a":
//
var extension = normalizedMime switch
{
"audio/m4a" or "audio/mp4" or "audio/aac" or "audio/x-m4a" => ".m4a",
"audio/ogg" => ".ogg",
"audio/webm" => ".webm",
_ => ".tmp"
};
tempFilePath = Path.Combine(
Path.GetTempPath(),
$"audio_{Guid.NewGuid()}{extension}"
);
//
stream.Position = 0;
using (var fileStream = File.Create(tempFilePath))
{
stream.CopyTo(fileStream);
}
//
MediaFoundationApi.Startup();
return new MediaFoundationReader(tempFilePath);
default:
//
tempFilePath = Path.Combine(
Path.GetTempPath(),
$"audio_{Guid.NewGuid()}.tmp"
);
stream.Position = 0;
using (var fileStream = File.Create(tempFilePath))
{
stream.CopyTo(fileStream);
}
try
{
MediaFoundationApi.Startup();
return new MediaFoundationReader(tempFilePath);
}
catch
{
return null;
}
}
}
catch (Exception)
{
return null;
}
}
#endregion
}
}