Use espeak-ng for coqui-ai/TTS VITS English models. (#466)

android/SherpaOnnxTts/app/src/main/java/com/k2fsa/sherpa/onnx/Tts.kt

@@ -23,7 +23,7 @@ data class OfflineTtsModelConfig(
 data class OfflineTtsConfig(
     var model: OfflineTtsModelConfig,
     var ruleFsts: String = "",
-    var maxNumSentences: Int = 2,
+    var maxNumSentences: Int = 1,
 )
 
 class GeneratedAudio(

python-api-examples/offline-tts-play.py

@@ -311,6 +311,9 @@ def main():
 
     if len(audio.samples) == 0:
         print("Error in generating audios. Please read previous error messages.")
+        global killed
+        killed = True
+        play_back_thread.join()
         return
 
     elapsed_seconds = end - start

scripts/apk/generate-tts-apk-script.py

@@ -33,6 +33,23 @@ class TtsModel:
     data_dir: Optional[str] = None
 
 
+def get_coqui_models() -> List[TtsModel]:
+    # English (coqui-ai/TTS)
+    models = [
+        TtsModel(model_dir="vits-coqui-en-ljspeech"),
+        TtsModel(model_dir="vits-coqui-en-ljspeech-neon"),
+        TtsModel(model_dir="vits-coqui-en-vctk"),
+        #  TtsModel(model_dir="vits-coqui-en-jenny"),
+    ]
+
+    for m in models:
+        m.data_dir = m.model_dir + "/" + "espeak-ng-data"
+        m.model_name = "model.onnx"
+        m.lang = "en"
+
+    return models
+
+
 def get_piper_models() -> List[TtsModel]:
     models = [
         TtsModel(model_dir="vits-piper-ar_JO-kareem-low"),
@@ -137,6 +154,7 @@ def get_piper_models() -> List[TtsModel]:
         TtsModel(model_dir="vits-piper-vi_VN-vivos-x_low"),
         TtsModel(model_dir="vits-piper-zh_CN-huayan-medium"),
     ]
+
     for m in models:
         m.data_dir = m.model_dir + "/" + "espeak-ng-data"
         m.model_name = m.model_dir[len("vits-piper-") :] + ".onnx"
@@ -145,7 +163,7 @@ def get_piper_models() -> List[TtsModel]:
     return models
 
 
-def get_all_models() -> List[TtsModel]:
+def get_vits_models() -> List[TtsModel]:
     return [
         # Chinese
         TtsModel(
@@ -202,12 +220,6 @@ def get_all_models() -> List[TtsModel]:
             lang="zh",
             rule_fsts="vits-zh-hf-theresa/rule.fst",
         ),
-        # English (coqui-ai/TTS)
-        # fmt: off
-        TtsModel(model_dir="vits-coqui-en-ljspeech", model_name="model.onnx", lang="en"),
-        TtsModel(model_dir="vits-coqui-en-ljspeech-neon", model_name="model.onnx", lang="en"),
-        TtsModel(model_dir="vits-coqui-en-vctk", model_name="model.onnx", lang="en"),
-        #  TtsModel(model_dir="vits-coqui-en-jenny", model_name="model.onnx", lang="en"),
         # English (US)
         TtsModel(model_dir="vits-vctk", model_name="vits-vctk.onnx", lang="en"),
         TtsModel(model_dir="vits-ljs", model_name="vits-ljs.onnx", lang="en"),
@@ -225,8 +237,11 @@ def main():
         s = f.read()
     template = environment.from_string(s)
     d = dict()
-    #  all_model_list = get_all_models()
+
+    #  all_model_list = get_vits_models()
     all_model_list = get_piper_models()
+    all_model_list += get_coqui_models()
+
     num_models = len(all_model_list)
 
     num_per_runner = num_models // total

sherpa-onnx/csrc/offline-tts-vits-impl.h

@@ -69,12 +69,16 @@ class OfflineTtsVitsImpl : public OfflineTtsImpl {
   }
 #endif
 
-  int32_t SampleRate() const override { return model_->SampleRate(); }
+  int32_t SampleRate() const override {
+    return model_->GetMetaData().sample_rate;
+  }
 
   GeneratedAudio Generate(
       const std::string &_text, int64_t sid = 0, float speed = 1.0,
       GeneratedAudioCallback callback = nullptr) const override {
-    int32_t num_speakers = model_->NumSpeakers();
+    const auto &meta_data = model_->GetMetaData();
+    int32_t num_speakers = meta_data.num_speakers;
+
     if (num_speakers == 0 && sid != 0) {
       SHERPA_ONNX_LOGE(
           "This is a single-speaker model and supports only sid 0. Given sid: "
@@ -105,14 +109,14 @@ class OfflineTtsVitsImpl : public OfflineTtsImpl {
     }
 
     std::vector<std::vector<int64_t>> x =
-        frontend_->ConvertTextToTokenIds(text, model_->Voice());
+        frontend_->ConvertTextToTokenIds(text, meta_data.voice);
 
     if (x.empty() || (x.size() == 1 && x[0].empty())) {
       SHERPA_ONNX_LOGE("Failed to convert %s to token IDs", text.c_str());
       return {};
     }
 
-    if (model_->AddBlank() && config_.model.vits.data_dir.empty()) {
+    if (meta_data.add_blank && config_.model.vits.data_dir.empty()) {
       for (auto &k : x) {
         k = AddBlank(k);
       }
@@ -189,25 +193,33 @@ class OfflineTtsVitsImpl : public OfflineTtsImpl {
  private:
 #if __ANDROID_API__ >= 9
   void InitFrontend(AAssetManager *mgr) {
-    if (model_->IsPiper() && !config_.model.vits.data_dir.empty()) {
+    const auto &meta_data = model_->GetMetaData();
+
+    if ((meta_data.is_piper || meta_data.is_coqui) &&
+        !config_.model.vits.data_dir.empty()) {
       frontend_ = std::make_unique<PiperPhonemizeLexicon>(
-          mgr, config_.model.vits.tokens, config_.model.vits.data_dir);
+          mgr, config_.model.vits.tokens, config_.model.vits.data_dir,
+          meta_data);
     } else {
       frontend_ = std::make_unique<Lexicon>(
           mgr, config_.model.vits.lexicon, config_.model.vits.tokens,
-          model_->Punctuations(), model_->Language(), config_.model.debug);
+          meta_data.punctuations, meta_data.language, config_.model.debug);
     }
   }
 #endif
 
   void InitFrontend() {
-    if (model_->IsPiper() && !config_.model.vits.data_dir.empty()) {
+    const auto &meta_data = model_->GetMetaData();
+
+    if ((meta_data.is_piper || meta_data.is_coqui) &&
+        !config_.model.vits.data_dir.empty()) {
       frontend_ = std::make_unique<PiperPhonemizeLexicon>(
-          config_.model.vits.tokens, config_.model.vits.data_dir);
+          config_.model.vits.tokens, config_.model.vits.data_dir,
+          model_->GetMetaData());
     } else {
       frontend_ = std::make_unique<Lexicon>(
           config_.model.vits.lexicon, config_.model.vits.tokens,
-          model_->Punctuations(), model_->Language(), config_.model.debug);
+          meta_data.punctuations, meta_data.language, config_.model.debug);
     }
   }
 
@@ -256,7 +268,7 @@ class OfflineTtsVitsImpl : public OfflineTtsImpl {
     const float *p = audio.GetTensorData<float>();
 
     GeneratedAudio ans;
-    ans.sample_rate = model_->SampleRate();
+    ans.sample_rate = model_->GetMetaData().sample_rate;
     ans.samples = std::vector<float>(p, p + total);
     return ans;
   }

sherpa-onnx/csrc/offline-tts-vits-model-config.cc

@@ -46,7 +46,8 @@ bool OfflineTtsVitsModelConfig::Validate() const {
 
   if (data_dir.empty()) {
     if (lexicon.empty()) {
-      SHERPA_ONNX_LOGE("Please provide --vits-lexicon");
+      SHERPA_ONNX_LOGE(
+          "Please provide --vits-lexicon if you leave --vits-data-dir empty");
       return false;
     }
 

sherpa-onnx/csrc/offline-tts-vits-model-metadata.h

@@ -0,0 +1,34 @@
+// sherpa-onnx/csrc/offline-tts-vits-model-metadata.h
+//
+// Copyright (c)  2023  Xiaomi Corporation
+
+#ifndef SHERPA_ONNX_CSRC_OFFLINE_TTS_VITS_MODEL_METADATA_H_
+#define SHERPA_ONNX_CSRC_OFFLINE_TTS_VITS_MODEL_METADATA_H_
+
+#include <cstdint>
+#include <string>
+
+namespace sherpa_onnx {
+
+struct OfflineTtsVitsModelMetaData {
+  int32_t sample_rate;
+  int32_t add_blank = 0;
+  int32_t num_speakers = 0;
+
+  std::string punctuations;
+  std::string language;
+  std::string voice;
+
+  bool is_piper = false;
+  bool is_coqui = false;
+
+  // the following options are for models from coqui-ai/TTS
+  int32_t blank_id = 0;
+  int32_t bos_id = 0;
+  int32_t eos_id = 0;
+  int32_t use_eos_bos = 0;
+};
+
+}  // namespace sherpa_onnx
+
+#endif  // SHERPA_ONNX_CSRC_OFFLINE_TTS_VITS_MODEL_METADATA_H_

sherpa-onnx/csrc/offline-tts-vits-model.cc

@@ -38,22 +38,14 @@ class OfflineTtsVitsModel::Impl {
 #endif
 
   Ort::Value Run(Ort::Value x, int64_t sid, float speed) {
-    if (is_piper_) {
+    if (meta_data_.is_piper || meta_data_.is_coqui) {
-      return RunVitsPiper(std::move(x), sid, speed);
+      return RunVitsPiperOrCoqui(std::move(x), sid, speed);
     }
 
     return RunVits(std::move(x), sid, speed);
   }
 
-  int32_t SampleRate() const { return sample_rate_; }
+  const OfflineTtsVitsModelMetaData &GetMetaData() const { return meta_data_; }
-
-  bool AddBlank() const { return add_blank_; }
-
-  std::string Punctuations() const { return punctuations_; }
-  std::string Language() const { return language_; }
-  std::string Voice() const { return voice_; }
-  bool IsPiper() const { return is_piper_; }
-  int32_t NumSpeakers() const { return num_speakers_; }
 
  private:
   void Init(void *model_data, size_t model_data_length) {
@@ -70,27 +62,52 @@ class OfflineTtsVitsModel::Impl {
       std::ostringstream os;
       os << "---vits model---\n";
       PrintModelMetadata(os, meta_data);
+
+      os << "----------input names----------\n";
+      int32_t i = 0;
+      for (const auto &s : input_names_) {
+        os << i << " " << s << "\n";
+        ++i;
+      }
+      os << "----------output names----------\n";
+      i = 0;
+      for (const auto &s : output_names_) {
+        os << i << " " << s << "\n";
+        ++i;
+      }
+
       SHERPA_ONNX_LOGE("%s\n", os.str().c_str());
     }
 
     Ort::AllocatorWithDefaultOptions allocator;  // used in the macro below
-    SHERPA_ONNX_READ_META_DATA(sample_rate_, "sample_rate");
+    SHERPA_ONNX_READ_META_DATA(meta_data_.sample_rate, "sample_rate");
-    SHERPA_ONNX_READ_META_DATA_WITH_DEFAULT(add_blank_, "add_blank", 0);
+    SHERPA_ONNX_READ_META_DATA_WITH_DEFAULT(meta_data_.add_blank, "add_blank",
-    SHERPA_ONNX_READ_META_DATA(num_speakers_, "n_speakers");
+                                            0);
-    SHERPA_ONNX_READ_META_DATA_STR_WITH_DEFAULT(punctuations_, "punctuation",
+    SHERPA_ONNX_READ_META_DATA(meta_data_.num_speakers, "n_speakers");
-                                                "");
+    SHERPA_ONNX_READ_META_DATA_STR_WITH_DEFAULT(meta_data_.punctuations,
-    SHERPA_ONNX_READ_META_DATA_STR(language_, "language");
+                                                "punctuation", "");
-    SHERPA_ONNX_READ_META_DATA_STR_WITH_DEFAULT(voice_, "voice", "");
+    SHERPA_ONNX_READ_META_DATA_STR(meta_data_.language, "language");
+    SHERPA_ONNX_READ_META_DATA_STR_WITH_DEFAULT(meta_data_.voice, "voice", "");
+
+    SHERPA_ONNX_READ_META_DATA_WITH_DEFAULT(meta_data_.blank_id, "blank_id", 0);
+    SHERPA_ONNX_READ_META_DATA_WITH_DEFAULT(meta_data_.bos_id, "bos_id", 0);
+    SHERPA_ONNX_READ_META_DATA_WITH_DEFAULT(meta_data_.eos_id, "eos_id", 0);
+    SHERPA_ONNX_READ_META_DATA_WITH_DEFAULT(meta_data_.use_eos_bos,
+                                            "use_eos_bos", 0);
 
     std::string comment;
     SHERPA_ONNX_READ_META_DATA_STR(comment, "comment");
-    if (comment.find("piper") != std::string::npos ||
+
-        comment.find("coqui") != std::string::npos) {
+    if (comment.find("piper") != std::string::npos) {
-      is_piper_ = true;
+      meta_data_.is_piper = true;
+    }
+
+    if (comment.find("coqui") != std::string::npos) {
+      meta_data_.is_coqui = true;
     }
   }
 
-  Ort::Value RunVitsPiper(Ort::Value x, int64_t sid, float speed) {
+  Ort::Value RunVitsPiperOrCoqui(Ort::Value x, int64_t sid, float speed) {
     auto memory_info =
         Ort::MemoryInfo::CreateCpu(OrtDeviceAllocator, OrtMemTypeDefault);
 
@@ -213,14 +230,7 @@ class OfflineTtsVitsModel::Impl {
   std::vector<std::string> output_names_;
   std::vector<const char *> output_names_ptr_;
 
-  int32_t sample_rate_;
+  OfflineTtsVitsModelMetaData meta_data_;
-  int32_t add_blank_;
-  int32_t num_speakers_;
-  std::string punctuations_;
-  std::string language_;
-  std::string voice_;
-
-  bool is_piper_ = false;
 };
 
 OfflineTtsVitsModel::OfflineTtsVitsModel(const OfflineTtsModelConfig &config)
@@ -239,21 +249,8 @@ Ort::Value OfflineTtsVitsModel::Run(Ort::Value x, int64_t sid /*=0*/,
   return impl_->Run(std::move(x), sid, speed);
 }
 
-int32_t OfflineTtsVitsModel::SampleRate() const { return impl_->SampleRate(); }
+const OfflineTtsVitsModelMetaData &OfflineTtsVitsModel::GetMetaData() const {
-
+  return impl_->GetMetaData();
-bool OfflineTtsVitsModel::AddBlank() const { return impl_->AddBlank(); }
-
-std::string OfflineTtsVitsModel::Punctuations() const {
-  return impl_->Punctuations();
-}
-
-std::string OfflineTtsVitsModel::Language() const { return impl_->Language(); }
-std::string OfflineTtsVitsModel::Voice() const { return impl_->Voice(); }
-
-bool OfflineTtsVitsModel::IsPiper() const { return impl_->IsPiper(); }
-
-int32_t OfflineTtsVitsModel::NumSpeakers() const {
-  return impl_->NumSpeakers();
 }
 
 }  // namespace sherpa_onnx

sherpa-onnx/csrc/offline-tts-vits-model.h

@@ -15,6 +15,7 @@
 
 #include "onnxruntime_cxx_api.h"  // NOLINT
 #include "sherpa-onnx/csrc/offline-tts-model-config.h"
+#include "sherpa-onnx/csrc/offline-tts-vits-model-metadata.h"
 
 namespace sherpa_onnx {
 
@@ -39,17 +40,7 @@ class OfflineTtsVitsModel {
    */
   Ort::Value Run(Ort::Value x, int64_t sid = 0, float speed = 1.0);
 
-  // Sample rate of the generated audio
+  const OfflineTtsVitsModelMetaData &GetMetaData() const;
-  int32_t SampleRate() const;
-
-  // true to insert a blank between each token
-  bool AddBlank() const;
-
-  std::string Punctuations() const;
-  std::string Language() const;  // e.g., Chinese, English, German, etc.
-  std::string Voice() const;     // e.g., en-us, for espeak-ng
-  bool IsPiper() const;
-  int32_t NumSpeakers() const;
 
  private:
   class Impl;

sherpa-onnx/csrc/piper-phonemize-lexicon.cc

@@ -57,10 +57,17 @@ static std::unordered_map<char32_t, int32_t> ReadTokens(std::istream &is) {
 
     s = conv.from_bytes(sym);
     if (s.size() != 1) {
+      // for tokens.txt from coqui-ai/TTS, the last token is <BLNK>
+      if (s.size() == 6 && s[0] == '<' && s[1] == 'B' && s[2] == 'L' &&
+          s[3] == 'N' && s[4] == 'K' && s[5] == '>') {
+        continue;
+      }
+
       SHERPA_ONNX_LOGE("Error when reading tokens at Line %s. size: %d",
                        line.c_str(), static_cast<int32_t>(s.size()));
       exit(-1);
     }
+
     char32_t c = s[0];
 
     if (token2id.count(c)) {
@@ -77,7 +84,7 @@ static std::unordered_map<char32_t, int32_t> ReadTokens(std::istream &is) {
 
 // see the function "phonemes_to_ids" from
 // https://github.com/rhasspy/piper/blob/master/notebooks/piper_inference_(ONNX).ipynb
-static std::vector<int64_t> PhonemesToIds(
+static std::vector<int64_t> PiperPhonemesToIds(
     const std::unordered_map<char32_t, int32_t> &token2id,
     const std::vector<piper::Phoneme> &phonemes) {
   // see
@@ -104,6 +111,65 @@ static std::vector<int64_t> PhonemesToIds(
   return ans;
 }
 
+static std::vector<int64_t> CoquiPhonemesToIds(
+    const std::unordered_map<char32_t, int32_t> &token2id,
+    const std::vector<piper::Phoneme> &phonemes,
+    const OfflineTtsVitsModelMetaData &meta_data) {
+  // see
+  // https://github.com/coqui-ai/TTS/blob/dev/TTS/tts/utils/text/tokenizer.py#L87
+  int32_t use_eos_bos = meta_data.use_eos_bos;
+  int32_t bos_id = meta_data.bos_id;
+  int32_t eos_id = meta_data.eos_id;
+  int32_t blank_id = meta_data.blank_id;
+  int32_t add_blank = meta_data.add_blank;
+  int32_t comma_id = token2id.at(',');
+  SHERPA_ONNX_LOGE("comma id: %d", comma_id);
+
+  std::vector<int64_t> ans;
+  if (add_blank) {
+    ans.reserve(phonemes.size() * 2 + 3);
+  } else {
+    ans.reserve(phonemes.size() + 2);
+  }
+
+  if (use_eos_bos) {
+    ans.push_back(bos_id);
+  }
+
+  if (add_blank) {
+    ans.push_back(blank_id);
+
+    for (auto p : phonemes) {
+      if (token2id.count(p)) {
+        ans.push_back(token2id.at(p));
+        ans.push_back(blank_id);
+      } else {
+        SHERPA_ONNX_LOGE("Skip unknown phonemes. Unicode codepoint: \\U+%04x.",
+                         static_cast<uint32_t>(p));
+      }
+    }
+  } else {
+    // not adding blank
+    for (auto p : phonemes) {
+      if (token2id.count(p)) {
+        ans.push_back(token2id.at(p));
+      } else {
+        SHERPA_ONNX_LOGE("Skip unknown phonemes. Unicode codepoint: \\U+%04x.",
+                         static_cast<uint32_t>(p));
+      }
+    }
+  }
+
+  // add a comma at the end of a sentence so that we can have a longer pause.
+  ans.push_back(comma_id);
+
+  if (use_eos_bos) {
+    ans.push_back(eos_id);
+  }
+
+  return ans;
+}
+
 void InitEspeak(const std::string &data_dir) {
   static std::once_flag init_flag;
   std::call_once(init_flag, [data_dir]() {
@@ -119,21 +185,23 @@ void InitEspeak(const std::string &data_dir) {
   });
 }
 
-PiperPhonemizeLexicon::PiperPhonemizeLexicon(const std::string &tokens,
+PiperPhonemizeLexicon::PiperPhonemizeLexicon(
-                                             const std::string &data_dir)
+    const std::string &tokens, const std::string &data_dir,
-    : data_dir_(data_dir) {
+    const OfflineTtsVitsModelMetaData &meta_data)
+    : meta_data_(meta_data) {
   {
     std::ifstream is(tokens);
     token2id_ = ReadTokens(is);
   }
 
-  InitEspeak(data_dir_);
+  InitEspeak(data_dir);
 }
 
 #if __ANDROID_API__ >= 9
-PiperPhonemizeLexicon::PiperPhonemizeLexicon(AAssetManager *mgr,
+PiperPhonemizeLexicon::PiperPhonemizeLexicon(
-                                             const std::string &tokens,
+    AAssetManager *mgr, const std::string &tokens, const std::string &data_dir,
-                                             const std::string &data_dir) {
+    const OfflineTtsVitsModelMetaData &meta_data)
+    : meta_data_(meta_data) {
   {
     auto buf = ReadFile(mgr, tokens);
     std::istrstream is(buf.data(), buf.size());
@@ -141,8 +209,9 @@ PiperPhonemizeLexicon::PiperPhonemizeLexicon(AAssetManager *mgr,
   }
 
   // We should copy the directory of espeak-ng-data from the asset to
-  // some internal or external storage and then pass the directory to data_dir.
+  // some internal or external storage and then pass the directory to
-  InitEspeak(data_dir_);
+  // data_dir.
+  InitEspeak(data_dir);
 }
 #endif
 
@@ -160,9 +229,21 @@ std::vector<std::vector<int64_t>> PiperPhonemizeLexicon::ConvertTextToTokenIds(
   std::vector<std::vector<int64_t>> ans;
 
   std::vector<int64_t> phoneme_ids;
-  for (const auto &p : phonemes) {
+
-    phoneme_ids = PhonemesToIds(token2id_, p);
+  if (meta_data_.is_piper) {
-    ans.push_back(std::move(phoneme_ids));
+    for (const auto &p : phonemes) {
+      phoneme_ids = PiperPhonemesToIds(token2id_, p);
+      ans.push_back(std::move(phoneme_ids));
+    }
+  } else if (meta_data_.is_coqui) {
+    for (const auto &p : phonemes) {
+      phoneme_ids = CoquiPhonemesToIds(token2id_, p, meta_data_);
+      ans.push_back(std::move(phoneme_ids));
+    }
+
+  } else {
+    SHERPA_ONNX_LOGE("Unsupported model");
+    exit(-1);
   }
 
   return ans;

sherpa-onnx/csrc/piper-phonemize-lexicon.h

@@ -15,25 +15,28 @@
 #endif
 
 #include "sherpa-onnx/csrc/offline-tts-frontend.h"
+#include "sherpa-onnx/csrc/offline-tts-vits-model-metadata.h"
 
 namespace sherpa_onnx {
 
 class PiperPhonemizeLexicon : public OfflineTtsFrontend {
  public:
-  PiperPhonemizeLexicon(const std::string &tokens, const std::string &data_dir);
+  PiperPhonemizeLexicon(const std::string &tokens, const std::string &data_dir,
+                        const OfflineTtsVitsModelMetaData &meta_data);
 
 #if __ANDROID_API__ >= 9
   PiperPhonemizeLexicon(AAssetManager *mgr, const std::string &tokens,
-                        const std::string &data_dir);
+                        const std::string &data_dir,
+                        const OfflineTtsVitsModelMetaData &meta_data);
 #endif
 
   std::vector<std::vector<int64_t>> ConvertTextToTokenIds(
       const std::string &text, const std::string &voice = "") const override;
 
  private:
-  std::string data_dir_;
   // map unicode codepoint to an integer ID
   std::unordered_map<char32_t, int32_t> token2id_;
+  OfflineTtsVitsModelMetaData meta_data_;
 };
 
 }  // namespace sherpa_onnx