EngineX-Iluvatar/enginex_bi_series-sherpa-onnx
Archived
16
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

Support Spanish in TTS (#396)

Browse Source
This commit is contained in:
Fangjun Kuang
2023-10-28 11:09:34 +08:00
committed by GitHub
parent 69e985f701
commit 64ab1ea9f8
3 changed files with 43 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

39
sherpa-onnx/csrc/text-utils.cc
View File

@@ -164,7 +164,7 @@ template bool SplitStringToFloats(const std::string &full, const char *delim,
std::vector<double> *out);
static bool IsPunct(char c) { return c != '\'' && std::ispunct(c); }
static bool IsGermanUmlauts(const std::string &words) {
static bool IsGermanUmlauts(const std::string &word) {
// ä 0xC3 0xA4
// ö 0xC3 0xB6
// ü 0xC3 0xBC
@@ -173,12 +173,12 @@ static bool IsGermanUmlauts(const std::string &words) {
// Ü 0xC3 0x9C
// ß 0xC3 0x9F
if (words.size() != 2 || static_cast<uint8_t>(words[0]) != 0xc3) {
if (word.size() != 2 || static_cast<uint8_t>(word[0]) != 0xc3) {
return false;
}
auto c = static_cast<uint8_t>(words[1]);
if (c == 0xa4 || c == 0xb6 || c == 0xbC || c == 0x84 || c == 0x96 ||
auto c = static_cast<uint8_t>(word[1]);
if (c == 0xa4 || c == 0xb6 || c == 0xbc || c == 0x84 || c == 0x96 ||
c == 0x9c || c == 0x9f) {
return true;
}
@@ -186,6 +186,33 @@ static bool IsGermanUmlauts(const std::string &words) {
return false;
}
// see https://www.tandem.net/blog/spanish-accents
static bool IsSpanishDiacritic(const std::string &word) {
// á 0xC3 0xA1
// é 0xC3 0xA9
// í 0xC3 0xAD
// ó 0xC3 0xB3
// ú 0xC3 0xBA
// ü 0xC3 0xBC
// ñ 0xC3 0xB1
if (word.size() != 2 || static_cast<uint8_t>(word[0]) != 0xc3) {
return false;
}
auto c = static_cast<uint8_t>(word[1]);
if (c == 0xa1 || c == 0xa9 || c == 0xad || c == 0xb3 || c == 0xba ||
c == 0xbc || c == 0xb1) {
return true;
}
return false;
}
static bool IsSpecial(const std::string &w) {
return IsGermanUmlauts(w) || IsSpanishDiacritic(w);
}
static std::vector<std::string> MergeCharactersIntoWords(
const std::vector<std::string> &words) {
std::vector<std::string> ans;
@@ -196,7 +223,7 @@ static std::vector<std::string> MergeCharactersIntoWords(
while (i < n) {
const auto &w = words[i];
if (w.size() >= 3 || (w.size() == 2 && !IsGermanUmlauts(w)) ||
if (w.size() >= 3 || (w.size() == 2 && !IsSpecial(w)) ||
(w.size() == 1 && (IsPunct(w[0]) || std::isspace(w[0])))) {
if (prev != -1) {
std::string t;
@@ -215,7 +242,7 @@ static std::vector<std::string> MergeCharactersIntoWords(
}
// e.g., öffnen
if (w.size() == 1 || (w.size() == 2 && IsGermanUmlauts(w))) {
if (w.size() == 1 || (w.size() == 2 && IsSpecial(w))) {
if (prev == -1) {
prev = i;
}