From 534c9d7736c711389af1a35ba9bb141366d55df2 Mon Sep 17 00:00:00 2001
From: Stephen Hodgson <hodgson.designs@gmail.com>
Date: Thu, 26 Oct 2023 14:16:04 -0400
Subject: [PATCH] add partial clip callback

---
 .../TextToSpeech/TextToSpeechEndpoint.cs      | 24 +++++++++++++++++--
 ElevenLabs-DotNet/VoiceClip.cs                |  4 ++--
 2 files changed, 24 insertions(+), 4 deletions(-)

diff --git a/ElevenLabs-DotNet/TextToSpeech/TextToSpeechEndpoint.cs b/ElevenLabs-DotNet/TextToSpeech/TextToSpeechEndpoint.cs
index f0a1bc2..31564d1 100644
--- a/ElevenLabs-DotNet/TextToSpeech/TextToSpeechEndpoint.cs
+++ b/ElevenLabs-DotNet/TextToSpeech/TextToSpeechEndpoint.cs
@@ -48,9 +48,13 @@ public TextToSpeechEndpoint(ElevenLabsClient api) : base(api) { }
         /// 4 - max latency optimizations, but also with text normalizer turned off for even more latency savings
         /// (best latency, but can mispronounce eg numbers and dates).
         /// </param>
+        /// <param name="partialClipCallback">
+        /// Optional, Callback to enable streaming audio as it comes in.<br/>
+        /// Returns partial <see cref="VoiceClip"/>s who's <see cref="VoiceClip.ClipData"/> overwritten with the next chunk of data.
+        /// </param>
         /// <param name="cancellationToken">Optional, <see cref="CancellationToken"/>.</param>
         /// <returns><see cref="VoiceClip"/>.</returns>
-        public async Task<VoiceClip> TextToSpeechAsync(string text, Voice voice, VoiceSettings voiceSettings = null, Model model = null, OutputFormat outputFormat = OutputFormat.MP3_44100_128, int? optimizeStreamingLatency = null, CancellationToken cancellationToken = default)
+        public async Task<VoiceClip> TextToSpeechAsync(string text, Voice voice, VoiceSettings voiceSettings = null, Model model = null, OutputFormat outputFormat = OutputFormat.MP3_44100_128, int? optimizeStreamingLatency = null, Func<VoiceClip, Task> partialClipCallback = null, CancellationToken cancellationToken = default)
         {
             if (text.Length > 5000)
             {
@@ -90,7 +94,23 @@ public async Task<VoiceClip> TextToSpeechAsync(string text, Voice voice, VoiceSe
 
             try
             {
-                await responseStream.CopyToAsync(memoryStream, cancellationToken).ConfigureAwait(false);
+                if (partialClipCallback == null)
+                {
+                    await responseStream.CopyToAsync(memoryStream, cancellationToken).ConfigureAwait(false);
+                }
+                else
+                {
+                    int bytesRead;
+                    var buffer = new byte[8192];
+
+                    while ((bytesRead = await responseStream.ReadAsync(buffer, cancellationToken).ConfigureAwait(false)) > 0)
+                    {
+                        var segment = new ReadOnlyMemory<byte>(buffer, 0, bytesRead);
+                        await partialClipCallback(new VoiceClip(clipId, text, voice, segment)).ConfigureAwait(false);
+                        await memoryStream.WriteAsync(segment, cancellationToken).ConfigureAwait(false);
+                    }
+                }
+
                 clipData = memoryStream.ToArray();
             }
             finally
diff --git a/ElevenLabs-DotNet/VoiceClip.cs b/ElevenLabs-DotNet/VoiceClip.cs
index 27c020a..3f76866 100644
--- a/ElevenLabs-DotNet/VoiceClip.cs
+++ b/ElevenLabs-DotNet/VoiceClip.cs
@@ -8,13 +8,13 @@ namespace ElevenLabs
 {
     public sealed class VoiceClip
     {
-        internal VoiceClip(string id, string text, Voice voice, byte[] clipData)
+        internal VoiceClip(string id, string text, Voice voice, ReadOnlyMemory<byte> clipData)
         {
             Id = id;
             Text = text;
             Voice = voice;
             TextHash = $"{id}{text}".GenerateGuid().ToString();
-            ClipData = new ReadOnlyMemory<byte>(clipData);
+            ClipData = clipData;
         }
 
         public string Id { get; }