Merge pull request #52 from OUCC/feat/#51

音声合成の生成時に長文を分割するように変更
OUCC · May 3, 2024 · dd4a5d2 · dd4a5d2
2 parents 372cf8c + 5c96b1e
commit dd4a5d2
Show file tree

Hide file tree

Showing 13 changed files with 117 additions and 41 deletions.
diff --git a/Epub/KoeBook.Epub/Contracts/Services/IScraperSelectorService.cs b/Epub/KoeBook.Epub/Contracts/Services/IScraperSelectorService.cs
@@ -12,5 +12,5 @@ public interface IScraperSelectorService
     /// </summary>
     public bool IsMatchSites(string url);
 
-    public ValueTask<EpubDocument> ScrapingAsync(string url, string coverFillePath, string tempDirectory, Guid id, CancellationToken ct);
+    public ValueTask<EpubDocument> ScrapingAsync(string url, string tempDirectory, Guid id, CancellationToken ct);
 }
diff --git a/Epub/KoeBook.Epub/Contracts/Services/IScrapingService.cs b/Epub/KoeBook.Epub/Contracts/Services/IScrapingService.cs
@@ -6,5 +6,5 @@ public interface IScrapingService
 {
     public bool IsMatchSite(Uri url);
 
-    public ValueTask<EpubDocument> ScrapingAsync(string url, string coverFillePath, string tempDirectory, Guid id, CancellationToken ct);
+    public ValueTask<EpubDocument> ScrapingAsync(string url, string tempDirectory, Guid id, CancellationToken ct);
 }
diff --git a/Epub/KoeBook.Epub/Models/EpubDocument.cs b/Epub/KoeBook.Epub/Models/EpubDocument.cs
@@ -4,7 +4,7 @@
 
 namespace KoeBook.Epub.Models;
 
-public class EpubDocument(string title, string author, string coverFilePath, Guid id)
+public class EpubDocument(string title, string author, Guid id, string coverFilePath = "")
 {
     public string Title { get; set; } = title;
     public string Author { get; set; } = author;

diff --git a/Epub/KoeBook.Epub/Services/AiStoryAnalyzerService.cs b/Epub/KoeBook.Epub/Services/AiStoryAnalyzerService.cs
@@ -11,7 +11,7 @@ public partial class AiStoryAnalyzerService(ISplitBraceService splitBraceService
 
     public EpubDocument CreateEpubDocument(AiStory aiStory, Guid id)
     {
-        return new EpubDocument(aiStory.Title, "AI", "", id)
+        return new EpubDocument(aiStory.Title, "AI", id)
         {
             Chapters = [new Chapter()
             {

diff --git a/Epub/KoeBook.Epub/Services/AnalyzerService.cs b/Epub/KoeBook.Epub/Services/AnalyzerService.cs
@@ -34,7 +34,7 @@ public async ValueTask<BookScripts> AnalyzeAsync(BookProperties bookProperties,
             switch (bookProperties)
             {
                 case { SourceType: SourceType.Url or SourceType.FilePath, Source: string uri }:
-                    document = await _scrapingService.ScrapingAsync(uri, coverFilePath, tempDirectory, bookProperties.Id, cancellationToken);
+                    document = await _scrapingService.ScrapingAsync(uri, tempDirectory, bookProperties.Id, cancellationToken);
                     break;
                 case { SourceType: SourceType.AiStory, Source: AiStory aiStory }:
                     document = _aiStoryAnalyzerService.CreateEpubDocument(aiStory, bookProperties.Id);
@@ -45,6 +45,7 @@ public async ValueTask<BookScripts> AnalyzeAsync(BookProperties bookProperties,
             }
 
             _createCoverFileService.Create(document.Title, document.Author, coverFilePath);
+            document.CoverFilePath = coverFilePath;
         }
         catch (EbookException) { throw; }
         catch (Exception ex)

diff --git a/Epub/KoeBook.Epub/Services/EpubGenerateService.cs b/Epub/KoeBook.Epub/Services/EpubGenerateService.cs
@@ -3,6 +3,7 @@
 using KoeBook.Core.Models;
 using KoeBook.Epub.Contracts.Services;
 using KoeBook.Epub.Models;
+using NAudio.Wave;
 
 namespace KoeBook.Epub.Services;
 
@@ -18,9 +19,15 @@ public async ValueTask<string> GenerateEpubAsync(BookScripts bookScripts, string
 
         var document = _documentStoreService.Documents.Single(d => d.Id == bookScripts.BookProperties.Id);
 
-        foreach (var scriptLine in bookScripts.ScriptLines)
+        for (var i = 0; i < bookScripts.ScriptLines.Length; i++)
         {
-            scriptLine.Audio = new Audio(await _soundGenerationService.GenerateLineSoundAsync(scriptLine, bookScripts.Options, cancellationToken).ConfigureAwait(false));
+            var scriptLine = bookScripts.ScriptLines[i];
+            var wavData = await _soundGenerationService.GenerateLineSoundAsync(scriptLine, bookScripts.Options, cancellationToken).ConfigureAwait(false);
+            using var ms = new MemoryStream(wavData);
+            using var reader = new WaveFileReader(ms);
+            var tmpMp3Path = Path.Combine(tempDirectory, $"{document.Title}{i}.mp3");
+            MediaFoundationEncoder.EncodeToMp3(reader, tmpMp3Path);
+            scriptLine.Audio = new Audio(reader.TotalTime, tmpMp3Path);
         }
 
         if (await _createService.TryCreateEpubAsync(document, tempDirectory, cancellationToken).ConfigureAwait(false))

diff --git a/Epub/KoeBook.Epub/Services/ScraperSelectorService.cs b/Epub/KoeBook.Epub/Services/ScraperSelectorService.cs
@@ -21,14 +21,14 @@ public bool IsMatchSites(string url)
         }
     }
 
-    public async ValueTask<EpubDocument> ScrapingAsync(string url, string coverFillePath, string tempDirectory, Guid id, CancellationToken ct)
+    public async ValueTask<EpubDocument> ScrapingAsync(string url, string tempDirectory, Guid id, CancellationToken ct)
     {
         var uri = new Uri(url);
 
         foreach (var service in _scrapingServices)
         {
             if (service.IsMatchSite(uri))
-                return await service.ScrapingAsync(url, coverFillePath, tempDirectory, id, ct);
+                return await service.ScrapingAsync(url, tempDirectory, id, ct);
         }
 
         throw new ArgumentException("対応するURLではありません");

diff --git a/Epub/KoeBook.Epub/Services/ScrapingAozoraService.cs b/Epub/KoeBook.Epub/Services/ScrapingAozoraService.cs
@@ -22,7 +22,7 @@ public bool IsMatchSite(Uri uri)
             return uri.Host == "www.aozora.gr.jp";
         }
 
-        public async ValueTask<EpubDocument> ScrapingAsync(string url, string coverFilePath, string imageDirectory, Guid id, CancellationToken ct)
+        public async ValueTask<EpubDocument> ScrapingAsync(string url, string imageDirectory, Guid id, CancellationToken ct)
         {
             var config = Configuration.Default.WithDefaultLoader();
             using var context = BrowsingContext.New(config);
@@ -37,7 +37,7 @@ public async ValueTask<EpubDocument> ScrapingAsync(string url, string coverFileP
                 ?? throw new EbookException(ExceptionType.WebScrapingFailed, $"著者の取得に失敗しました。\n以下のリンクから正しい小説のリンクを取得してください。\n{GetCardUrl(url)}");
 
             // EpubDocument の生成
-            var document = new EpubDocument(TextReplace(bookTitle.InnerHtml), TextReplace(bookAuther.InnerHtml), coverFilePath, id);
+            var document = new EpubDocument(TextReplace(bookTitle.InnerHtml), TextReplace(bookAuther.InnerHtml), id);
 
             var (contentsIds, hasChapter, hasSection) = LoadToc(doc, document);
 

diff --git a/Epub/KoeBook.Epub/Services/ScrapingNaroService.cs b/Epub/KoeBook.Epub/Services/ScrapingNaroService.cs
@@ -25,7 +25,7 @@ public bool IsMatchSite(Uri uri)
             return uri.Host == "ncode.syosetu.com";
         }
 
-        public async ValueTask<EpubDocument> ScrapingAsync(string url, string coverFilePath, string imageDirectory, Guid id, CancellationToken ct)
+        public async ValueTask<EpubDocument> ScrapingAsync(string url, string imageDirectory, Guid id, CancellationToken ct)
         {
             var ncode = GetNcode(url);
             var novelInfo = await GetNovelInfoAsync(ncode, ct).ConfigureAwait(false);
@@ -53,7 +53,7 @@ public async ValueTask<EpubDocument> ScrapingAsync(string url, string coverFileP
                 ? bookAuthorTag.InnerHtml
                 : bookAuthorElement.InnerHtml.Replace("作者：", "");
 
-            var document = new EpubDocument(bookTitle, bookAuthor, coverFilePath, id);
+            var document = new EpubDocument(bookTitle, bookAuthor, id);
             if (novelInfo.IsSerial) // 連載の時
             {
                 async IAsyncEnumerable<(string? title, Section section)> LoadDetailsAsync(IBrowsingContext context, NovelInfo novelInfo, string imageDirectory, [EnumeratorCancellation] CancellationToken ct)

diff --git a/KoeBook.Core/Models/Audio.cs b/KoeBook.Core/Models/Audio.cs
@@ -1,25 +1,15 @@
-using NAudio.Wave;
+using System.IO;
+using NAudio.Wave;
 
 namespace KoeBook.Epub.Models;
 
-public sealed class Audio
+public sealed class Audio(TimeSpan totalTIme, string tempFilePath)
 {
-    public TimeSpan TotalTime { get; }
-    private readonly byte[] _mp3Data;
+    public TimeSpan TotalTime { get; } = totalTIme;
+    public string TempFilePath { get; } = tempFilePath;
 
-    public Audio(byte[] mp3Data)
+    public FileStream GetStream()
     {
-        _mp3Data = mp3Data;
-        using var ms = new MemoryStream();
-        ms.Write(_mp3Data.AsSpan());
-        ms.Flush();
-        ms.Position = 0;
-        using var reader = new Mp3FileReader(ms);
-        TotalTime = reader.TotalTime;
-    }
-
-    public MemoryStream GetStream()
-    {
-        return new MemoryStream(_mp3Data);
+        return new FileStream(TempFilePath, FileMode.Open, FileAccess.Read, FileShare.Read, 4096, true);
     }
 }
diff --git a/KoeBook.Core/Services/SoundGenerationService.cs b/KoeBook.Core/Services/SoundGenerationService.cs
@@ -1,6 +1,10 @@
-using System.Web;
+using System.Buffers;
+using System.Runtime.CompilerServices;
+using System.Threading;
+using System.Web;
 using KoeBook.Core.Contracts.Services;
 using KoeBook.Core.Models;
+using NAudio.Wave;
 
 namespace KoeBook.Core.Services;
 
@@ -17,11 +21,85 @@ public async ValueTask<byte[]> GenerateLineSoundAsync(ScriptLine scriptLine, Boo
         var soundModel = _soundGenerationSelectorService.Models.FirstOrDefault(m => m.Name == model)
             ?? throw new EbookException(ExceptionType.SoundGenerationFailed);
         var style = soundModel.Styles.Contains(scriptLine.Style) ? scriptLine.Style : soundModel.Styles[0];
-        var queryCollection = HttpUtility.ParseQueryString(string.Empty);
-        queryCollection.Add("text", scriptLine.Text);
-        queryCollection.Add("model_id", soundModel.Id);
-        queryCollection.Add("style", style);
-        return await _styleBertVitsClientService
-            .GetAsByteArrayAsync($"/voice?{queryCollection}", ExceptionType.SoundGenerationFailed, cancellationToken).ConfigureAwait(false);
+        using var msWriter = new MemoryStream();
+        WaveFileWriter? writer = null;
+        byte[] dataBuffer = ArrayPool<byte>.Shared.Rent(1024);
+        try
+        {
+            await foreach (var voice in GenerateSoundAsync(scriptLine.Text, style, soundModel.Id, cancellationToken))
+            {
+                if (voice.Length > dataBuffer.Length)
+                {
+                    ArrayPool<byte>.Shared.Return(dataBuffer);
+                    dataBuffer = ArrayPool<byte>.Shared.Rent(voice.Length);
+                }
+                using var msReader = new MemoryStream(voice);
+                using var reader = new WaveFileReader(msReader);
+                var read = await reader.ReadAsync(dataBuffer, cancellationToken);
+                if (writer is null)
+                {
+                    writer = new WaveFileWriter(msWriter, reader.WaveFormat);
+                }
+                await writer.WriteAsync(dataBuffer.AsMemory()[..read], cancellationToken);
+            }
+            if (writer is null)
+            {
+                throw new EbookException(ExceptionType.SoundGenerationFailed);
+            }
+            await writer.FlushAsync(cancellationToken);
+            return msWriter.ToArray();
+        }
+        catch { throw; }
+        finally
+        {
+            ArrayPool<byte>.Shared?.Return(dataBuffer);
+            writer?.Dispose();
+        }
+    }
+
+    private async IAsyncEnumerable<byte[]> GenerateSoundAsync(string text, string style, string modelId, [EnumeratorCancellation] CancellationToken cancellationToken)
+    {
+        foreach (var l in SplitPeriod(text, 300))
+        {
+            var queryCollection = HttpUtility.ParseQueryString(string.Empty);
+            queryCollection.Add("text", l);
+            queryCollection.Add("model_id", modelId);
+            queryCollection.Add("style", style);
+            yield return await _styleBertVitsClientService
+                .GetAsByteArrayAsync($"/voice?{queryCollection}", ExceptionType.SoundGenerationFailed, cancellationToken).ConfigureAwait(false);
+        }
+    }
+
+    private IEnumerable<string> SplitPeriod(string text, int limit)
+    {
+        if (text.Length < limit)
+        {
+            yield return text;
+        }
+        else
+        {
+            List<int> periodList = [0];
+            var textSpan = text.AsSpan();
+            var chunk = textSpan[..limit];
+            while (true)
+            {
+                var periodIndex = periodList[^1] + chunk.LastIndexOf('。') + 1;
+                periodList.Add(periodIndex);
+                var nextEnd = periodIndex + limit;
+                if (nextEnd < textSpan.Length)
+                {
+                    chunk = textSpan[periodIndex..nextEnd];
+                }
+                else
+                {
+                    periodList.Add(textSpan.Length);
+                    break;
+                }
+            }
+            for (var i = 1; i < periodList.Count; i++)
+            {
+                yield return text[periodList[i - 1]..periodList[i]];
+            }
+        }
     }
 }
diff --git a/KoeBook.Test/Epub/EpubDocumentTest.cs b/KoeBook.Test/Epub/EpubDocumentTest.cs
@@ -8,7 +8,7 @@ public class EpubDocumentTest
     [Fact]
     public void EnsureChapter()
     {
-        var document = new EpubDocument("title", "author", "cover", default);
+        var document = new EpubDocument("title", "author", default);
 
         Assert.Empty(document.Chapters);
 
@@ -29,7 +29,7 @@ public void EnsureChapter()
     [Fact]
     public void EnsureSection()
     {
-        var document = new EpubDocument("title", "author", "cover", default);
+        var document = new EpubDocument("title", "author", default);
 
         Assert.Empty(document.Chapters);
 
@@ -77,7 +77,7 @@ public void EnsureSection()
     [Fact]
     public void EnsureParagraph()
     {
-        var document = new EpubDocument("title", "author", "cover", default);
+        var document = new EpubDocument("title", "author", default);
 
         Assert.Empty(document.Chapters);
 

diff --git a/KoeBook.Test/Epub/ScrapingAozoraServiceTest.cs b/KoeBook.Test/Epub/ScrapingAozoraServiceTest.cs
@@ -28,7 +28,7 @@ public async Task AddParagraphs1(string input, string[] expected)
         using var context = BrowsingContext.New(Configuration.Default);
         using var doc = await context.OpenAsync(req => req.Content(input));
         Assert.NotNull(doc.ParentElement);
-        var epubDocument = new EpubDocument("title", "author", "", default)
+        var epubDocument = new EpubDocument("title", "author", default)
         {
             Chapters = [new() { Sections = [new("section title") { Elements = [new Paragraph() { Text = "test" }] }] }]
         };