Add arguments

- language: specify transcribe language
- only extract: only extract audio

ConsoleWhisper/Model/Argument.cs

@@ -5,20 +5,28 @@
 namespace ConsoleWhisper.Model {
 	public class Argument {
 		[Option('i', "input", Required = true, Hidden = false, Separator = ' ', HelpText = "Input media files.")]
-		[Value(3)]
+		[Value(4)]
 		public IEnumerable<string> Files { get; set; }
 
 		[Option('m', "model", Required = false, Hidden = false, Default = "small", HelpText = "Whisper model: base, tiny, small, medium, large.")]
 		[Value(0, Min = 0, Max = 1, Required = false)]
 		public string ModelType { get; set; }
 
-		[Option('o', "output", Required = false, Hidden = false, HelpText = "Output directory.")]
+		[Option('o', "output", Required = false, Hidden = false, HelpText = "(Default: current directory) Output directory.")]
 		[Value(1, Min = 0, Max = 1, Required = false)]
 		public string OutputDir { get; set; }
 
-		[Option('g', "gpu", Required = false, Hidden = true, HelpText = "Currently not implemented.")]
-		[Value(2, Required = false, Default = false)]
-		public bool GPU { get; set; }
+		[Option('l', "language", Required = false, Hidden = false, Default = "auto", HelpText = "Specify transcribe language.")]
+		[Value(2, Required = false)]
+		public string Language { get; set; }
+
+		[Option("only-extract", Required = false, Hidden = false, Default = false, HelpText = "Extract audio stream, without transcribing.")]
+		[Value(3, Required = false)]
+		public bool OnlyExtract { get; set; }
+
+		//[Option('g', "gpu", Required = false, Hidden = true, HelpText = "Currently not implemented.")]
+		//[Value(2, Required = false, Default = false)]
+		//public bool GPU { get; set; }
 
 		public Argument() {
 			OutputDir = Environment.CurrentDirectory;
@@ -27,10 +35,20 @@ public class Argument {
 		public void Validate() {
 			if (!SupportedModels.Contains(ModelType))
 				throw new ArgumentException(message: $"Whisper model type \"{ModelType}\" is not supported.");
+			if (!SupportedLanguages.Contains(Language))
+				throw new ArgumentException(message: $"Language \"{Language}\" is not supported.\nCheck {LanguageLink} for available languages.");
 		}
 
-		internal const int SupportedArgumentsCount = 5;
+		internal const int SupportedArgumentsCount = 7;
 
 		private static readonly HashSet<string> SupportedModels = new() { "base", "tiny", "small", "medium", "large" };
+		private static readonly HashSet<string> SupportedLanguages = new() { "en", "zh", "de", "es", "ru", "ko", "fr", "ja", "pt", "tr",
+			"pl", "ca", "nl", "ar", "sv", "it", "id", "hi", "fi", "vi", "he", "uk", "el", "ms", "ro", "da", "hu", "ta", "no", "th", "ur",
+			"hr", "bg", "lt", "la", "mi", "ml", "cy", "sk", "te", "fa", "lv", "bn", "sr", "az", "sl", "kn", "et", "mk", "br", "eu", "is",
+			"hy", "ne", "mn", "bs", "kk", "sq", "sw", "gl", "mr", "pa", "si", "km", "sn", "yo", "so", "af", "oc", "ka", "be", "tg", "sd",
+			"gu", "am", "yi", "lo", "uz", "fo", "ht", "ps", "tk", "nn", "mt", "sa", "lb", "my", "bo", "tl", "mg", "as", "tt", "haw", "ln",
+			"ha", "ba", "jw", "su", "auto" };
+
+		private const string LanguageLink = "https://github.com/ggerganov/whisper.cpp/blob/57543c169e27312e7546d07ed0d8c6eb806ebc36/whisper.cpp#L121";
 	}
 }

ConsoleWhisper/Module/AudioHelper.cs

@@ -1,15 +1,16 @@
 using NAudio.Wave;
 using System;
 using System.Collections.Generic;
+using System.IO;
 using System.Linq;
 using System.Threading.Tasks;
 using Xabe.FFmpeg;
 
 namespace ConsoleWhisper.Module {
 	public static class AudioHelper {
-		public static async Task<string> Extract(string mediaFilename) {
+		public static async Task<string> Extract(string outputDir, string mediaFilename, bool isOnlyExtract) {
 			try {
-				var audioFilename = FileHelper.GetTempMp3File();
+				var audioFilename = isOnlyExtract ? FileHelper.GetAudioPath(outputDir, mediaFilename) : FileHelper.GetTempMp3File();
 
 				var mediaInfo = await FFmpeg.GetMediaInfo(mediaFilename);
 
@@ -20,10 +21,14 @@ public static class AudioHelper {
 
 				await DoConversion(audioStream, audioFilename);
 
-				var resampledWaveFilename = Resample(audioFilename);
-				FileHelper.DelFile(audioFilename);
+				if (!isOnlyExtract) {
+					var resampledWaveFilename = Resample(audioFilename);
+					FileHelper.DelFile(audioFilename);
 
-				return resampledWaveFilename;
+					return resampledWaveFilename;
+				}
+
+				return audioFilename;
 			} catch (Exception) {
 				throw;
 			}
@@ -36,6 +41,8 @@ public static class AudioHelper {
 				.SetOutputFormat(Format.mp3)
 				.SetOutput(audioFilename);
 
+			Output.Info($"Extracting audio to {Path.GetFileName(audioFilename)}");
+
 			await conversion.Start();
 		}
 

ConsoleWhisper/Module/FileHelper.cs

@@ -35,6 +35,21 @@ internal static class FileHelper {
 			File.Delete(filePath);
 		}
 
+		internal static string GetTranscriptPath(string outputDir, string filename) {
+			var extension = Path.GetExtension(filename).TrimStart('.');
+			var filenameWithoutExtension = Path.GetFileNameWithoutExtension(filename);
+			var transcriptName = Path.ChangeExtension($"{filenameWithoutExtension}-{extension}", SrtExtension);
+			return Path.Combine(outputDir, transcriptName);
+		}
+
+		internal static string GetAudioPath(string outputDir, string filename) {
+			var extension = Path.GetExtension(filename).TrimStart('.');
+			var filenameWithoutExtension = Path.GetFileNameWithoutExtension(filename);
+			var transcriptName = Path.ChangeExtension($"{filenameWithoutExtension}-{extension}", Mp3Extension);
+			return Path.Combine(outputDir, transcriptName);
+		}
+
+		#region Get temp file name
 		internal static string GetTempFile() {
 			return Path.GetTempFileName();
 		}
@@ -59,7 +74,9 @@ internal static class FileHelper {
 			DelFile(tempFilename);
 			return waveFilename;
 		}
+		#endregion
 
+		#region Whisper model related operations
 		internal static bool ModelExists(string modelFilename) { 
 			return File.Exists(Path.Combine(ModelDirectory, modelFilename));
 		}
@@ -71,14 +88,9 @@ internal static class FileHelper {
 		internal static string GetModelPath(string modelType) {
 			return Path.Combine(ModelDirectory, modelType);
 		}
+		#endregion
 
-		internal static string GetTranscriptPath(string outputDir, string filename) {
-			var extension = Path.GetExtension(filename).TrimStart('.');
-			var filenameWithoutExtension = Path.GetFileNameWithoutExtension(filename);
-			var transcriptName = Path.ChangeExtension($"{filenameWithoutExtension}-{extension}", SrtExtension);
-			return Path.Combine(outputDir, transcriptName);
-		}
-
+		#region Add Text to Filestream
 		internal static async Task AddText(FileStream fs, string value) {
 			byte[] info = encoder.GetBytes(value);
 			await fs.WriteAsync(info);
@@ -93,12 +105,15 @@ internal static class FileHelper {
 			byte[] info = encoder.GetBytes(value.ToString("G"));
 			await fs.WriteAsync(info);
 		}
+		#endregion
 
 		private static readonly UTF8Encoding encoder = new(true);
 
+		#region Extension string
 		private const string WaveExtension = "wav";
 		private const string AacExtension = "aac";
 		private const string Mp3Extension = "mp3";
 		private const string SrtExtension = "srt";
+		#endregion
 	}
 }

ConsoleWhisper/Module/WhisperHelper.cs

@@ -7,12 +7,12 @@
 
 namespace ConsoleWhisper.Module {
 	internal class WhisperHelper {
-		internal static async Task Transcribe(string modelType, string wavFilename, string mediaFilename, string outputDir) {
+		internal static async Task Transcribe(string modelType, string wavFilename, string mediaFilename, string outputDir, string language) {
 			try {
 				using var whisperFactory = WhisperFactory.FromPath(FileHelper.GetModelPath(modelType));
 
 				using var processor = whisperFactory.CreateBuilder()
-					.WithLanguage("auto")
+					.WithLanguage(language)
 					.WithPrintProgress()
 					.Build();
 

ConsoleWhisper/Program.cs

@@ -48,10 +48,13 @@ internal class Program {
 				int cnt = 1;
 				foreach (var file in arg.Files) {
 					var mediaFilename = Path.GetFileName(file);
-					var wavFilename = await AudioHelper.Extract(file);
-					Output.Info($"Start transcribing file #{cnt++}: {mediaFilename}");
-					await WhisperHelper.Transcribe(arg.ModelType, wavFilename, mediaFilename, arg.OutputDir);
-					FileHelper.DelFile(wavFilename);
+					var wavFilename = await AudioHelper.Extract(arg.OutputDir, file, arg.OnlyExtract);
+
+					if (!arg.OnlyExtract) {
+						Output.Info($"Start transcribing file #{cnt++}: {mediaFilename}");
+						await WhisperHelper.Transcribe(arg.ModelType, wavFilename, mediaFilename, arg.OutputDir, arg.Language);
+						FileHelper.DelFile(wavFilename);
+					}
 				}