Add support for whisper large v3 turbo

Also:

* Double # of audio device slots
* Fetch CuDNN from NVIDIA at runtime instead of vendoring

4 files changed, 169 insertions, 101 deletions

diff --git a/GUI/GUI/GUI/Frame.cpp b/GUI/GUI/GUI/Frame.cpp
index 77447bc..0782376 100644
--- a/GUI/GUI/GUI/Frame.cpp
+++ b/GUI/GUI/GUI/Frame.cpp
@@ -136,6 +136,16 @@ namespace {
         "7",

         "8",

         "9",

+        "10",

+        "11",

+        "12",

+        "13",

+        "14",

+        "15",

+        "16",

+        "17",

+        "18",

+        "19",

     };

     const size_t kNumMicChoices = sizeof(kMicChoices) / sizeof(kMicChoices[0]);

     constexpr int kMicDefault = 0;  // index

@@ -472,6 +482,7 @@ namespace {
         "large-v2",

         "distil-large-v3",

         "large-v3",

+        "large-v3-turbo",

     };

     const size_t kNumModelChoices = sizeof(kModelChoices) / sizeof(kModelChoices[0]);

     constexpr int kModelDefault = 2;  // base.en

@@ -1855,6 +1866,27 @@ void Frame::OnNavbarDebug(wxCommandEvent& event)
     Resize();

 }

 

+bool FindDirectoryByPrefix(wxTextCtrl* out, const std::string& prefix, std::filesystem::path& path) {

+	std::error_code ec;

+	// Find directory starting with "cudnn"

+	for (const auto& entry :

+		std::filesystem::directory_iterator(".", ec)) {

+		if (ec) {

+			Log(out, "Failed to iterate cwd: {}\n",

+				ec.message());

+			return false;

+		}

+

+		if (entry.is_directory(ec) &&

+			!ec &&

+			entry.path().filename().string().starts_with(prefix)) {

+			path = entry.path();

+            return true;

+		}

+	}

+    return false;

+}

+

 void Frame::EnsureVirtualEnv(bool block, bool force)

 {

     auto status = env_proc_.wait_for(std::chrono::seconds(0));

@@ -1886,48 +1918,116 @@ void Frame::EnsureVirtualEnv(bool block, bool force)
         return;

 	}

 

-    env_proc_ = std::move(std::async(std::launch::async, [&]() {

-        Log(transcribe_out_, "Setting up Python virtual environment\n");

-		Log(transcribe_out_, "This could take several minutes, please be "

-			"patient!\n");

-		Log(transcribe_out_, "This will download ~1GB of dependencies.\n");

-

-		{

-			Log(transcribe_out_, "  Installing pip\n");

-			auto out_cb = [&](const std::string& out, const std::string& err) {

-				Log(transcribe_out_, "{}", out);

-				Log(transcribe_out_, "{}", err);

-			};

-			if (!PythonWrapper::InstallPip(std::move(out_cb))) {

-                Log(transcribe_out_, "Failed to install pip!\n");

-                return false;

-			}

-		}

+	env_proc_ = std::move(std::async(std::launch::async, [&]() {

+		Log(transcribe_out_, "Setting up Python virtual environment\n");

+	Log(transcribe_out_, "This could take several minutes, please be "

+		"patient!\n");

+	Log(transcribe_out_, "This will download ~1GB of dependencies.\n");

 

-		Log(transcribe_out_, "  Installing dependencies\n");

+	{

+		Log(transcribe_out_, "  Installing pip\n");

 		auto out_cb = [&](const std::string& out, const std::string& err) {

 			Log(transcribe_out_, "{}", out);

 			Log(transcribe_out_, "{}", err);

 		};

-		if (!PythonWrapper::InvokeWithArgs(*app_c_,

-			{

-				"-u",  // Unbuffered output

-				"-m pip",

-				"install",

-				"-r Resources/Scripts/requirements_frozen.txt",

-			}, std::move(out_cb))) {

-			Log(transcribe_out_, "Failed to launch environment setup thread!\n");

+		if (!PythonWrapper::InstallPip(std::move(out_cb))) {

+			Log(transcribe_out_, "Failed to install pip!\n");

 			return false;

 		}

-        Log(transcribe_out_, "Successfully set up virtual environment!\n");

+	}

 

-        std::ofstream venv_flag_ofs(venv_flag);

-		auto now = std::chrono::system_clock::now();

-		const int64_t seconds_since_epoch = std::chrono::duration_cast<std::chrono::seconds>(now.time_since_epoch()).count();

-        venv_flag_ofs << std::to_string(seconds_since_epoch);

+	Log(transcribe_out_, "  Installing python dependencies\n");

+	auto out_cb = [&](const std::string& out, const std::string& err) {

+		Log(transcribe_out_, "{}", out);

+		Log(transcribe_out_, "{}", err);

+	};

+	if (!PythonWrapper::InvokeWithArgs(*app_c_,

+		{

+			"-u",  // Unbuffered output

+			"-m pip",

+			"install",

+			"-r Resources/Scripts/requirements_frozen.txt",

+		}, out_cb)) {

+		Log(transcribe_out_, "Failed to launch environment setup thread!\n");

+		return false;

+	}

 

-		return true;

-	}));

+	Log(transcribe_out_, " Fetching CuDNN\n");

+	if (!PythonWrapper::InvokeWithArgs(*app_c_,

+		{

+			"-m wget",

+			"https://developer.download.nvidia.com/compute/cudnn/redist/cudnn/windows-x86_64/cudnn-windows-x86_64-9.5.1.17_cuda12-archive.zip"

+		}, out_cb)) {

+	}

+	Log(transcribe_out_, " Unzipping CuDNN\n");

+	if (!PythonWrapper::InvokeWithArgs(*app_c_,

+		{

+			"-m zipfile -e",

+			"cudnn-windows-x86_64-9.5.1.17_cuda12-archive.zip",

+			"."

+		}, out_cb)) {

+	}

+	Log(transcribe_out_, " Installing CuDNN\n");

+	{

+		std::filesystem::path cudnn_dir;

+        if (!FindDirectoryByPrefix(transcribe_out_, "cudnn", cudnn_dir)) {

+			Log(transcribe_out_, "Failed to find unzipped cudnn "

+				"directory\n");

+            return false;

+        }

+		std::error_code ec;

+		const std::filesystem::path dest_dir = "Resources/Scripts";

+		for (const auto& entry :

+			std::filesystem::recursive_directory_iterator(cudnn_dir, ec)) {

+			if (ec) {

+				Log(transcribe_out_, "Failed to iterate cudnn dir: {}\n",

+					ec.message());

+				return false;

+			}

+			if (!entry.is_regular_file(ec)) {

+				continue;

+			}

+			if (ec) {

+				Log(transcribe_out_, "Skipping unrecognized file type "

+					"{}: {}\n", entry.path().string(), ec.message());

+				continue;

+			}

+			if (entry.path().extension() != ".dll") {

+				continue;

+			}

+			std::filesystem::path dest =

+				dest_dir / entry.path().filename();

+			// Remove destination

+			std::filesystem::remove(dest, ec);

+			if (ec) {

+				Log(transcribe_out_, "Failed to delete old CuDNN .dll: {}\n",

+					ec.message());

+				return false;

+			}

+			// Rename file

+			std::filesystem::rename(entry.path(), dest, ec);

+			if (ec) {

+				Log(transcribe_out_, "Failed to move CuDNN .dll: {}\n",

+					ec.message());

+			}

+		}

+		// Delete cudnn dir

+        std::filesystem::remove_all(cudnn_dir, ec);

+        if (ec) {

+            Log(transcribe_out_, "Failed to remove old CuDNN dir: {}\n",

+                ec.message());

+        }

+	}

+

+	Log(transcribe_out_, "Successfully set up virtual environment!\n");

+

+	std::ofstream venv_flag_ofs(venv_flag);

+	auto now = std::chrono::system_clock::now();

+	const int64_t seconds_since_epoch = std::chrono::duration_cast<std::chrono::seconds>(now.time_since_epoch()).count();

+	venv_flag_ofs << std::to_string(seconds_since_epoch);

+

+	return true;

+		}));

 

     if (block) {

         // Spinning prevents the GUI from hanging.

diff --git a/GUI/package.ps1 b/GUI/package.ps1
index 9b1c05b..697f1a2 100644
--- a/GUI/package.ps1
+++ b/GUI/package.ps1
@@ -85,39 +85,6 @@ if (-Not (Test-Path $git_dir)) {
   Read-Host -Prompt "Press enter once PortableGit is installed at $pwd\PortableGit"

 }

 

-$nvidia_dir = "nvidia_dll"

-

-if (-Not (Test-Path $nvidia_dir)) {

-  echo "Fetching CUDNN dependencies"

-

-  mkdir $nvidia_dir

-  pushd $nvidia_dir > $null

-

-  $ZLIB_URL = "https://drive.google.com/uc?export=download&id=1NpWU83JVOWG0tJtFK7ObygTbOasGWZpI"

-  Invoke-WebRequest $ZLIB_URL -OutFile "zlibwapi.dll"

-

-  # NVIDIA locks these files behind a fucking login making it a massive

-  # pain in the dick for end users to download, so I rehosted them.

-  # TODO check hashes.

-  echo "Fetching NVIDIA dll 1/4 (90 MB)"

-  $CUDNN_1_URL = "https://www.dropbox.com/scl/fi/d21dsoa982ce7wigng510/cudnn_ops_infer64_8.dll?rlkey=xflxyux0ekhr0fs11m4gs58md&st=0wff5fyn&dl=1"

-  Invoke-WebRequest $CUDNN_1_URL -OutFile "cudnn_ops_infer64_8.dll"

-

-  echo "Fetching NVIDIA dll 2/4 (570 MB)"

-  $CUDNN_2_URL = "https://www.dropbox.com/scl/fi/uqccevwk9h2q84dt9vr6u/cudnn_cnn_infer64_8.dll?rlkey=sik7xd0ozg06nr4eayzdym4la&st=031bb8pa&dl=1"

-  Invoke-WebRequest $CUDNN_2_URL -OutFile "cudnn_cnn_infer64_8.dll"

-

-  echo "Fetching NVIDIA dll 3/4 (470 MB)"

-  $CUBLAS_1_URL = "https://www.dropbox.com/scl/fi/3vrd4fzwno8q5ejigqz6l/cublasLt64_12.dll?rlkey=uvbdn5e7dmm8ajdhm7yztjmhc&st=dguf57q4&dl=1"

-  Invoke-WebRequest $CUBLAS_1_URL -OutFile "cublasLt64_12.dll"

-

-  echo "Fetching NVIDIA dll 4/4 (100 MB)"

-  $CUBLAS_2_URL = "https://www.dropbox.com/scl/fi/hoxjdru7qmwbzelw1gr2t/cublas64_12.dll?rlkey=mcmq5t0b62wjc2uc7ylrixwi6&st=z1la337w&dl=1"

-  Invoke-WebRequest $CUBLAS_2_URL -OutFile "cublas64_12.dll"

-

-  popd > $null

-}

-

 if (-Not (Test-Path UwwwuPP)) {

   git clone https://github.com/yum-food/UwwwuPP

   pushd UwwwuPP > $null

@@ -162,7 +129,6 @@ cp -Recurse ../Fonts/Emotes TaSTT/Resources/Fonts/Emotes
 cp -Recurse Python TaSTT/Resources/Python

 cp -Recurse PortableGit TaSTT/Resources/PortableGit

 cp -Recurse ../Scripts TaSTT/Resources/Scripts

-cp $nvidia_dir/*.dll TaSTT/Resources/Scripts/

 mkdir TaSTT/Resources/Images

 cp ../Images/logo*.png TaSTT/Resources/Images/

 cp -Recurse ../Shaders TaSTT/Resources/Shaders

diff --git a/Scripts/requirements.txt b/Scripts/requirements.txt
index 949b82c..41c581c 100644
--- a/Scripts/requirements.txt
+++ b/Scripts/requirements.txt
@@ -1,6 +1,6 @@
-ctranslate2==4.2.1
+ctranslate2==4.5.0
 editdistance
-faster-whisper@https://github.com/guillaumekln/faster-whisper/archive/2f6913efc85306fc4f900da6c67f9a06a7d54a3d.tar.gz
+faster-whisper@https://github.com/guillaumekln/faster-whisper/archive/53bbe5401683c9a7549db62642e3d4535956b95c.tar.gz
 future==0.18.2
 huggingface_hub==0.16.4
 keyboard
@@ -14,4 +14,5 @@ python-osc
 pyyaml
 sentence_splitter
 transformers>=4.21.0
+wget
 
diff --git a/Scripts/requirements_frozen.txt b/Scripts/requirements_frozen.txt
index 1913457..9e6a6ab 100644
--- a/Scripts/requirements_frozen.txt
+++ b/Scripts/requirements_frozen.txt
@@ -1,41 +1,42 @@
-av==12.0.0
-certifi==2023.7.22
-charset-normalizer==3.2.0
+av==13.1.0
+certifi==2024.8.30
+charset-normalizer==3.4.0
 colorama==0.4.6
 coloredlogs==15.0.1
-ctranslate2==4.2.1
-editdistance==0.6.2
-faster-whisper @ https://github.com/guillaumekln/faster-whisper/archive/2f6913efc85306fc4f900da6c67f9a06a7d54a3d.tar.gz#sha256=c389ad787c8cdafcb13d31f8bae788083eb4e490819aad6b49d76c82e490a388
-filelock==3.12.3
-flatbuffers==23.5.26
-fsspec==2023.9.0
+ctranslate2==4.5.0
+editdistance==0.8.1
+faster-whisper @ https://github.com/guillaumekln/faster-whisper/archive/53bbe5401683c9a7549db62642e3d4535956b95c.tar.gz#sha256=17b49d15a58e18d78b4639af59bd35da12bc0bf3bb73c9af4ad48891dd6793f7
+filelock==3.16.1
+flatbuffers==24.3.25
+fsspec==2024.10.0
 future==0.18.2
 huggingface-hub==0.16.4
 humanfriendly==10.0
-idna==3.4
+idna==3.10
 keyboard==0.13.5
-langcodes==3.3.0
-language-data==1.1
-marisa-trie==0.7.8
+langcodes==3.4.1
+language_data==1.2.0
+marisa-trie==1.2.1
 mpmath==1.3.0
-numpy==1.25.2
-onnxruntime==1.14.1
-openvr==1.26.701
-packaging==23.1
-Pillow==10.0.0
-protobuf==4.24.3
-PyAudio==0.2.13
+numpy==2.1.3
+onnxruntime==1.20.0
+openvr==2.5.101
+packaging==24.2
+pillow==11.0.0
+protobuf==5.28.3
+PyAudio==0.2.14
 pydub==0.25.1
-pyreadline3==3.4.1
-python-osc==1.8.3
-PyYAML==6.0.1
-regex==2023.8.8
-requests==2.31.0
-safetensors==0.3.3
+pyreadline3==3.5.4
+python-osc==1.9.0
+PyYAML==6.0.2
+regex==2024.11.6
+requests==2.32.3
+safetensors==0.4.5
 sentence-splitter==1.4
-sympy==1.12
-tokenizers==0.13.3
-tqdm==4.66.1
-transformers==4.33.1
-typing_extensions==4.7.1
-urllib3==2.0.4
+sympy==1.13.3
+tokenizers==0.15.2
+tqdm==4.67.0
+transformers==4.35.2
+typing_extensions==4.12.2
+urllib3==2.2.3
+wget==3.2