Fix Edge online speech with Sec-MS-GEC v2

NaturalVoiceSAPIAdapter/EdgeGec.h

@@ -0,0 +1,145 @@
+#pragma once
+#include <stdint.h>
+#include <stdio.h>
+#include <string.h>
+
+// Sec-MS-GEC v2, ported from edge-tts 7.3.0 drm.py (LGPL-3.0).
+// Keep this independent of Windows/SAPI so the production algorithm can be tested.
+namespace EdgeGec {
+#include "EdgeGecTables.h"
+inline uint32_t Ror(uint32_t x, unsigned n) {
+    return (x >> n) | (x << (32 - n));
+}
+inline uint32_t Swap(uint32_t x) {
+    return Ror(x, 16);
+}
+inline uint32_t S0(uint32_t x) {
+    return Ror(x, 7) ^ Ror(x, 18) ^ (x >> 3);
+}
+inline uint32_t S1(uint32_t x) {
+    return Ror(x, 17) ^ Ror(x, 19) ^ (x >> 10);
+}
+inline uint32_t ReadBE(const unsigned char* p) {
+    return uint32_t(p[0]) << 24 | uint32_t(p[1]) << 16 | uint32_t(p[2]) << 8 | p[3];
+}
+inline uint32_t ReadLE(const unsigned char* p) {
+    return uint32_t(p[3]) << 24 | uint32_t(p[2]) << 16 | uint32_t(p[1]) << 8 | p[0];
+}
+inline void Expand(uint32_t* w, unsigned first) {
+    for (unsigned i = first; i < 64; i++) w[i] = S1(w[i - 2]) + w[i - 7] + S0(w[i - 15]) + w[i - 16];
+}
+inline void Compress(uint32_t* state, const uint32_t* w, unsigned first) {
+    uint32_t a = state[0], b = state[1], c = state[2], d = state[3], e = state[4], f = state[5], g = state[6],
+             h = state[7];
+    for (unsigned i = first; i < 64; i++) {
+        uint32_t k = first == 14 && i == 14 ? 0xC9DC06A7 : Rounds[i];
+        uint32_t t1 = h + (Ror(e, 6) ^ Ror(e, 11) ^ Ror(e, 25)) + (g ^ (e & (f ^ g))) + k + w[i];
+        uint32_t t2 = (Ror(a, 2) ^ Ror(a, 13) ^ Ror(a, 22)) + ((a & b) | (c & (a | b)));
+        h = g;
+        g = f;
+        f = e;
+        e = d + t1;
+        d = c;
+        c = b;
+        b = a;
+        a = t1 + t2;
+    }
+    uint32_t work[] = {a, b, c, d, e, f, g, h};
+    for (unsigned i = 0; i < 8; i++) state[i] = (first == 14 ? Initial[i] : state[i]) + work[i];
+}
+// fileTime includes the correction derived from the server's Date header.
+// result must have room for 81 ASCII characters plus the terminating NUL.
+inline void Generate(uint64_t fileTime, char* result) {
+    fileTime -= fileTime % 3000000000ULL;
+    char data[64];
+    snprintf(data, sizeof(data), "%llu6A5AA1D4EAFF4E9FB37E23D68491D6F4", (unsigned long long)fileTime);
+    size_t len = strlen(data);
+    const auto* bytes = reinterpret_cast<const unsigned char*>(data);
+    // V2 resumes a modified SHA-256 block after a fixed 57-byte prefix.
+    // These schedule/state adjustments are intentional; ordinary SHA256(data)
+    // produces the old v1 token and cannot replace this step.
+    uint32_t w[64] = {};
+    w[14] = uint32_t(bytes[0]) << 16 | uint32_t(bytes[1]) << 8 | bytes[2];
+    w[15] = ReadBE(bytes + 3);
+    for (unsigned i = 16; i < 30; i++) {
+        uint32_t v = Schedule[i - 16];
+        if (i - 2 >= 14) v += S1(w[i - 2]) ^ (i == 16 ? 0x000B92C0 : 0);
+        if (i - 7 >= 14) v += w[i - 7];
+        if (i - 15 >= 14) v += S0(w[i - 15]) ^ (i == 29 ? 0x059C0B80 : 0);
+        w[i] = v;
+    }
+    w[30] = S1(w[28]) + w[23] + S0(w[15]) + w[14] + 0xC9DC06A7 - Rounds[14];
+    Expand(w, 31);
+    uint32_t state[8];
+    memcpy(state, ResumeState, sizeof(state));
+    Compress(state, w, 14);
+    unsigned char padded[128] = {};
+    memcpy(padded, bytes + 7, len - 7);
+    padded[len - 7] = 0x80;
+    size_t total = ((len + 57 + 9 + 63) / 64) * 64 - 64;
+    uint64_t bits = (len + 57) * 8;
+    for (unsigned i = 0; i < 8; i++) padded[total - 1 - i] = (unsigned char)(bits >> (i * 8));
+    for (size_t off = 0; off < total; off += 64) {
+        for (unsigned i = 0; i < 16; i++) w[i] = ReadBE(padded + off + i * 4);
+        Expand(w, 16);
+        Compress(state, w, 0);
+    }
+    unsigned char digest[32];
+    for (unsigned i = 0; i < 32; i++) digest[i] = (unsigned char)(state[i / 4] >> (24 - 8 * (i % 4)));
+    // CS64 uses little-endian words; SHA-256 emits big-endian words.
+    uint32_t ka = ReadLE(digest) | 1, kb = ReadLE(digest + 4) | 1;
+    uint32_t factors[] = {ka * 0x82893933U, kb * 0x82893933U};
+    uint32_t terms[32] = {}, offsets[32] = {};
+    unsigned count = 14;
+    for (unsigned i = 0; i < 14; i++) {
+        terms[i] = Coefficients[i][0] * factors[i % 2];
+        offsets[i] = Coefficients[i][1];
+    }
+    unsigned char input[72] = {};
+    memcpy(input + 1, bytes, len);
+    unsigned words = (unsigned)((len + 1 + 7) / 8) * 2;
+    for (unsigned i = 0; i < words; i++) {
+        uint32_t word = ReadLE(input + i * 4);
+        terms[count] = word * (i % 2 ? kb : ka);
+        offsets[count] = Swap(word) * (i % 2 ? 0xC31713DBU : 0xEF0569FBU);
+        if (i == 0) {
+            terms[count] += 0xF2F90B1AU * factors[0];
+            offsets[count] += 0x0B1A0000;
+        }
+        count++;
+    }
+    uint32_t swapAcc = 0, swapLast = 0, s = 0, acc = 0;
+    for (unsigned i = 0; i < count; i += 2) {
+        uint32_t x = terms[i], y = terms[i + 1];
+        uint32_t q = Swap(offsets[i] + x) * 0x76F50F34U;
+        swapAcc += Swap(q) * 0xC3EFEA97U + q;
+        q = Swap(offsets[i + 1] + y) * 0x1881CE3CU;
+        swapLast = Swap(q) * 0x35BD1EC9U + q;
+        swapAcc += swapLast;
+        s = s * ka + x;
+        s = Swap(s) * 0xCF98B111U;
+        s = Swap(s) * 0x87085B9FU;
+        s = Swap(s) * 0x12CEB96DU;
+        s = Swap(s) * 0x257E1D83U;
+        acc += s;
+        uint32_t t = s * kb + y;
+        t = Swap(t) * 0xA27416F5U;
+        t = Swap(t) * 0xD38396FFU;
+        s = Swap(Swap(t) * 0x7C932B89U) * 0xBFA49F69U;
+        acc += s;
+    }
+    uint32_t cs[] = {swapLast ^ s, swapAcc ^ acc};
+    constexpr char hex[] = "0123456789ABCDEF";
+    for (unsigned i = 0; i < 8; i++) {
+        unsigned char v = (unsigned char)(cs[i / 4] >> (8 * (i % 4)));
+        *result++ = hex[v >> 4];
+        *result++ = hex[v & 15];
+    }
+    *result++ = '-';
+    for (auto v : digest) {
+        *result++ = hex[v >> 4];
+        *result++ = hex[v & 15];
+    }
+    *result = 0;
+}
+} // namespace EdgeGec

NaturalVoiceSAPIAdapter/EdgeGecTables.h

@@ -0,0 +1,27 @@
+// Constants from edge-tts 7.3.0 drm.py (LGPL-3.0). See tests/reference.
+static constexpr uint32_t Initial[] = {
+    0x6A09E667, 0xBB67AE85, 0x3C6EF372, 0xA54FF53A, 0x510E527F, 0x9B05688C, 0x1F83D9AB, 0x5BE0CD19,
+};
+static constexpr uint32_t ResumeState[] = {
+    0xFD9FBEA2, 0x91243CE7, 0xBE499094, 0xCD2E1E0A, 0x6C006F49, 0x04BD35AB, 0xFEB0A7E4, 0x03EFFBD2,
+};
+static constexpr uint32_t Schedule[] = {
+    0xEB24BF9E, 0x1558691F, 0xA0C09EE2, 0x0CFA32BA, 0xA622EE8C, 0xD1DD9491, 0xC0DD25A0,
+    0xF9E70AB6, 0xF0C31573, 0x7058CE4E, 0x7188E606, 0x00E5AB33, 0xD258B060, 0x72766564,
+};
+static constexpr uint32_t Rounds[] = {
+    0x428A2F98, 0x71374491, 0xB5C0FBCF, 0xE9B5DBA5, 0x3956C25B, 0x59F111F1, 0x923F82A4, 0xAB1C5ED5,
+    0xD807AA98, 0x12835B01, 0x243185BE, 0x550C7DC3, 0x72BE5D74, 0x80DEB1FE, 0x9BDC06A7, 0xC19BF174,
+    0xE49B69C1, 0xEFBE4786, 0x0FC19DC6, 0x240CA1CC, 0x2DE92C6F, 0x4A7484AA, 0x5CB0A9DC, 0x76F988DA,
+    0x983E5152, 0xA831C66D, 0xB00327C8, 0xBF597FC7, 0xC6E00BF3, 0xD5A79147, 0x06CA6351, 0x14292967,
+    0x27B70A85, 0x2E1B2138, 0x4D2C6DFC, 0x53380D13, 0x650A7354, 0x766A0ABB, 0x81C2C92E, 0x92722C85,
+    0xA2BFE8A1, 0xA81A664B, 0xC24B8B70, 0xC76C51A3, 0xD192E819, 0xD6990624, 0xF40E3585, 0x106AA070,
+    0x19A4C116, 0x1E376C08, 0x2748774C, 0x34B0BCB5, 0x391C0CB3, 0x4ED8AA4A, 0x5B9CCA4F, 0x682E6FF3,
+    0x748F82EE, 0x78A5636F, 0x84C87814, 0x8CC70208, 0x90BEFFFA, 0xA4506CEB, 0xBEF9A3F7, 0xC67178F2,
+};
+static constexpr uint32_t Coefficients[][2] = {
+    {0xE82E91B1, 0xB32700D0}, {0x44DE24C6, 0x88DFF51D}, {0x3BEB65BC, 0x29514038}, {0xC89C5933, 0xB328CBDF},
+    {0x8D1EC211, 0xD405B4D5}, {0x314EF5D5, 0x26E55F3C}, {0xE92491B1, 0x607801C6}, {0xF0A5B9D5, 0x82B2C1FB},
+    {0x06064CF3, 0xFA51A3DA}, {0xDA79FA60, 0x207CC464}, {0xBED20560, 0x3C3C74F3}, {0x89F9C9F8, 0xD2D14B61},
+    {0x35FA38C6, 0xBB04A2C1}, {0xCB49E3C6, 0xC5626167},
+};

NaturalVoiceSAPIAdapter/NaturalVoiceSAPIAdapter.vcxproj

@@ -102,6 +102,9 @@
     <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
   </ImportGroup>
   <PropertyGroup Label="UserMacros" />
+  <PropertyGroup>
+    <NaturalVoiceRegisterOutput Condition="'$(NaturalVoiceRegisterOutput)' == ''">true</NaturalVoiceRegisterOutput>
+  </PropertyGroup>
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
     <IgnoreImportLibrary>true</IgnoreImportLibrary>
     <LinkIncremental>true</LinkIncremental>
@@ -174,7 +177,7 @@
     <Link>
       <SubSystem>Windows</SubSystem>
       <ModuleDefinitionFile>.\NaturalVoiceSAPIAdapter.def</ModuleDefinitionFile>
-      <RegisterOutput>true</RegisterOutput>
+      <RegisterOutput>$(NaturalVoiceRegisterOutput)</RegisterOutput>
       <PerUserRedirection>false</PerUserRedirection>
       <DelayLoadDLLs>Microsoft.CognitiveServices.Speech.core.dll</DelayLoadDLLs>
       <AdditionalDependencies>libcrypto.lib;libssl.lib;crypt32.lib;RuntimeObject.lib;%(AdditionalDependencies)</AdditionalDependencies>
@@ -214,7 +217,7 @@
     <Link>
       <SubSystem>Windows</SubSystem>
       <ModuleDefinitionFile>.\NaturalVoiceSAPIAdapter.def</ModuleDefinitionFile>
-      <RegisterOutput>true</RegisterOutput>
+      <RegisterOutput>$(NaturalVoiceRegisterOutput)</RegisterOutput>
       <PerUserRedirection>false</PerUserRedirection>
       <DelayLoadDLLs>Microsoft.CognitiveServices.Speech.core.dll</DelayLoadDLLs>
       <AdditionalDependencies>libcrypto.lib;libssl.lib;crypt32.lib;RuntimeObject.lib;%(AdditionalDependencies)</AdditionalDependencies>
@@ -255,7 +258,7 @@
     <Link>
       <SubSystem>Windows</SubSystem>
       <ModuleDefinitionFile>.\NaturalVoiceSAPIAdapter.def</ModuleDefinitionFile>
-      <RegisterOutput>true</RegisterOutput>
+      <RegisterOutput>$(NaturalVoiceRegisterOutput)</RegisterOutput>
       <PerUserRedirection>false</PerUserRedirection>
       <DelayLoadDLLs>Microsoft.CognitiveServices.Speech.core.dll</DelayLoadDLLs>
       <AdditionalDependencies>libcrypto.lib;libssl.lib;crypt32.lib;RuntimeObject.lib;%(AdditionalDependencies)</AdditionalDependencies>
@@ -298,7 +301,7 @@
       <ModuleDefinitionFile>.\NaturalVoiceSAPIAdapter.def</ModuleDefinitionFile>
       <EnableCOMDATFolding>true</EnableCOMDATFolding>
       <OptimizeReferences>true</OptimizeReferences>
-      <RegisterOutput>true</RegisterOutput>
+      <RegisterOutput>$(NaturalVoiceRegisterOutput)</RegisterOutput>
       <PerUserRedirection>false</PerUserRedirection>
       <DelayLoadDLLs>Microsoft.CognitiveServices.Speech.core.dll</DelayLoadDLLs>
       <AdditionalOptions>/PDBALTPATH:%_PDB% %(AdditionalOptions)</AdditionalOptions>
@@ -340,7 +343,7 @@
       <ModuleDefinitionFile>.\NaturalVoiceSAPIAdapter.def</ModuleDefinitionFile>
       <EnableCOMDATFolding>true</EnableCOMDATFolding>
       <OptimizeReferences>true</OptimizeReferences>
-      <RegisterOutput>true</RegisterOutput>
+      <RegisterOutput>$(NaturalVoiceRegisterOutput)</RegisterOutput>
       <PerUserRedirection>false</PerUserRedirection>
       <DelayLoadDLLs>Microsoft.CognitiveServices.Speech.core.dll</DelayLoadDLLs>
       <AdditionalOptions>/PDBALTPATH:%_PDB% %(AdditionalOptions)</AdditionalOptions>
@@ -383,7 +386,7 @@
       <ModuleDefinitionFile>.\NaturalVoiceSAPIAdapter.def</ModuleDefinitionFile>
       <EnableCOMDATFolding>true</EnableCOMDATFolding>
       <OptimizeReferences>true</OptimizeReferences>
-      <RegisterOutput>true</RegisterOutput>
+      <RegisterOutput>$(NaturalVoiceRegisterOutput)</RegisterOutput>
       <PerUserRedirection>false</PerUserRedirection>
       <DelayLoadDLLs>Microsoft.CognitiveServices.Speech.core.dll</DelayLoadDLLs>
       <AdditionalOptions>/PDBALTPATH:%_PDB% %(AdditionalOptions)</AdditionalOptions>
@@ -417,6 +420,8 @@
     <ClInclude Include="VoiceToken.h" />
     <ClInclude Include="VoiceTokenEnumerator.h" />
     <ClInclude Include="WSConnectionPool.h" />
+    <ClInclude Include="EdgeGec.h" />
+    <ClInclude Include="EdgeGecTables.h" />
     <ClInclude Include="WSLogger.h" />
     <ClInclude Include="wrappers.h" />
   </ItemGroup>

NaturalVoiceSAPIAdapter/SpeechServiceConstants.h

@@ -16,14 +16,16 @@ constexpr inline const char* EDGE_TRUSTED_CLIENT_TOKEN = "6A5AA1D4EAFF4E9FB37E23
 
 constexpr inline const char* EDGE_VOICE_LIST_URL = "https://speech.platform.bing.com/consumer/speech/synthesize/readaloud/voices/list?trustedclienttoken=6A5AA1D4EAFF4E9FB37E23D68491D6F4";
 
-constexpr inline const wchar_t* EDGE_WEBSOCKET_URL
-	= L"wss://api.msedgeservices.com/tts/cognitiveservices/websocket/v1"
-	  L"?Ocp-Apim-Subscription-Key=6A5AA1D4EAFF4E9FB37E23D68491D6F4&Sec-MS-GEC={Sec-MS-GEC}&Sec-MS-GEC-Version=1-142.0.3595.94";
+constexpr inline const wchar_t* EDGE_WEBSOCKET_URL =
+    L"wss://speech.platform.bing.com/consumer/speech/synthesize/readaloud/edge/v1"
+    L"?TrustedClientToken=6A5AA1D4EAFF4E9FB37E23D68491D6F4&Sec-MS-GEC={Sec-MS-GEC}&Sec-MS-GEC-Version=2-143.0.3650.75";
 
 constexpr inline const char* AZURE_TTS_HOST_AFTER_REGION = ".tts.speech.microsoft.com";
 
 constexpr inline const char* AZURE_VOICE_LIST_PATH = "/cognitiveservices/voices/list";
 
 constexpr inline const char* AZURE_WEBSOCKET_PATH = "/cognitiveservices/websocket/v1";
 
-constexpr inline const char* EDGE_USER_AGENT = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/142.0.0.0 Safari/537.36 Edg/142.0.0.0";
+constexpr inline const char* EDGE_USER_AGENT =
+    "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/143.0.0.0 Safari/537.36 "
+    "Edg/143.0.0.0";