add multi-speaker support for speedyspeech

examples/aishell3/tts2/default_multi.yaml

@@ -0,0 +1,52 @@
+###########################################################
+#                FEATURE EXTRACTION SETTING               #
+###########################################################
+fs: 24000           # Sampling rate.
+n_fft: 2048         # FFT size (samples).
+n_shift: 300        # Hop size (samples). 12.5ms
+win_length: 1200    # Window length (samples). 50ms
+                    # If set to null, it will be the same as fft_size.
+window: "hann"      # Window function.
+n_mels: 80          # Number of mel basis.
+fmin: 80            # Minimum freq in mel basis calculation.
+fmax: 7600          # Maximum frequency in mel basis calculation.
+
+###########################################################
+#                       DATA SETTING                      #
+###########################################################
+batch_size: 32
+num_workers: 4
+
+###########################################################
+#                       MODEL SETTING                     #
+###########################################################
+model:
+  encoder_hidden_size: 128
+  encoder_kernel_size: 3
+  encoder_dilations: [1, 3, 9, 27, 1, 3, 9, 27, 1, 1]
+  duration_predictor_hidden_size: 128
+  decoder_hidden_size: 128
+  decoder_output_size: 80
+  decoder_kernel_size: 3
+  decoder_dilations: [1, 3, 9, 27, 1, 3, 9, 27, 1, 3, 9, 27, 1, 3, 9, 27, 1, 1]
+  spk_embed_dim: 256   
+  spk_embed_integration_type: add         # speaker embedding integration type
+
+###########################################################
+#                     OPTIMIZER SETTING                   #
+###########################################################
+optimizer:
+  optim: adam              # optimizer type
+  learning_rate: 0.002     # learning rate
+  max_grad_norm: 1
+
+###########################################################
+#                     TRAINING SETTING                    #
+###########################################################
+max_epoch: 100
+num_snapshots: 5
+
+###########################################################
+#                       OTHER SETTING                     #
+###########################################################
+seed: 10086

paddlespeech/t2s/datasets/am_batch_fn.py

@@ -17,7 +17,7 @@ import paddle
 from paddlespeech.t2s.data.batch import batch_sequences
 
 
-def speedyspeech_batch_fn(examples):
+def speedyspeech_single_spk_batch_fn(examples):
     # fields = ["phones", "tones", "num_phones", "num_frames", "feats", "durations"]
     phones = [np.array(item["phones"], dtype=np.int64) for item in examples]
     tones = [np.array(item["tones"], dtype=np.int64) for item in examples]
@@ -54,6 +54,46 @@ def speedyspeech_batch_fn(examples):
     }
     return batch
 
+def speedyspeech_multi_spk_batch_fn(examples):
+    # fields = ["phones", "tones", "num_phones", "num_frames", "feats", "durations"]
+    phones = [np.array(item["phones"], dtype=np.int64) for item in examples]
+    tones = [np.array(item["tones"], dtype=np.int64) for item in examples]
+    feats = [np.array(item["feats"], dtype=np.float32) for item in examples]
+    durations = [
+        np.array(item["durations"], dtype=np.int64) for item in examples
+    ]
+    num_phones = [
+        np.array(item["num_phones"], dtype=np.int64) for item in examples
+    ]
+    num_frames = [
+        np.array(item["num_frames"], dtype=np.int64) for item in examples
+    ]
+
+    phones = batch_sequences(phones)
+    tones = batch_sequences(tones)
+    feats = batch_sequences(feats)
+    durations = batch_sequences(durations)
+
+    # convert each batch to paddle.Tensor
+    phones = paddle.to_tensor(phones)
+    tones = paddle.to_tensor(tones)
+    feats = paddle.to_tensor(feats)
+    durations = paddle.to_tensor(durations)
+    num_phones = paddle.to_tensor(num_phones)
+    num_frames = paddle.to_tensor(num_frames)
+    batch = {
+        "phones": phones,
+        "tones": tones,
+        "num_phones": num_phones,
+        "num_frames": num_frames,
+        "feats": feats,
+        "durations": durations,
+    }
+    if "spk_id" in examples[0]:
+        spk_id = [np.array(item["spk_id"], dtype=np.int64) for item in examples]
+        spk_id = paddle.to_tensor(spk_id)
+        batch["spk_id"] = spk_id
+    return batch
 
 def fastspeech2_single_spk_batch_fn(examples):
     # fields = ["text", "text_lengths", "speech", "speech_lengths", "durations", "pitch", "energy"]

paddlespeech/t2s/exps/speedyspeech/normalize.py

@@ -47,7 +47,8 @@ def main():
         "--phones-dict", type=str, default=None, help="phone vocabulary file.")
     parser.add_argument(
         "--tones-dict", type=str, default=None, help="tone vocabulary file.")
-
+    parser.add_argument(
+        "--speaker-dict", type=str, default=None, help="speaker id map file.")
     parser.add_argument(
         "--verbose",
         type=int,
@@ -121,6 +122,12 @@ def main():
     for tone, id in tone_id:
         vocab_tones[tone] = int(id)
 
+    vocab_speaker = {}
+    with open(args.speaker_dict, 'rt') as f:
+        spk_id = [line.strip().split() for line in f.readlines()]
+    for spk, id in spk_id:
+        vocab_speaker[spk] = int(id)
+
     # process each file
     output_metadata = []
 
@@ -135,11 +142,13 @@ def main():
         np.save(mel_path, mel.astype(np.float32), allow_pickle=False)
         phone_ids = [vocab_phones[p] for p in item['phones']]
         tone_ids = [vocab_tones[p] for p in item['tones']]
+        spk_id = vocab_speaker[item["speaker"]]
         if args.use_relative_path:
             # convert absolute path to relative path:
             mel_path = mel_path.relative_to(dumpdir)
         output_metadata.append({
             'utt_id': utt_id,
+            "spk_id": spk_id,
             'phones': phone_ids,
             'tones': tone_ids,
             'num_phones': item['num_phones'],

paddlespeech/t2s/exps/speedyspeech/preprocess.py

@@ -13,6 +13,7 @@
 # limitations under the License.
 import argparse
 import re
+import os
 from concurrent.futures import ThreadPoolExecutor
 from operator import itemgetter
 from pathlib import Path
@@ -32,7 +33,7 @@ from paddlespeech.t2s.datasets.preprocess_utils import compare_duration_and_mel_
 from paddlespeech.t2s.datasets.preprocess_utils import get_phn_dur
 from paddlespeech.t2s.datasets.preprocess_utils import get_phones_tones
 from paddlespeech.t2s.datasets.preprocess_utils import merge_silence
-
+from paddlespeech.t2s.datasets.preprocess_utils import get_spk_id_map
 
 def process_sentence(config: Dict[str, Any],
                      fp: Path,
@@ -101,6 +102,7 @@ def process_sentence(config: Dict[str, Any],
             "utt_id": utt_id,
             "phones": phones,
             "tones": tones,
+            "speaker": speaker,
             "num_phones": len(phones),
             "num_frames": num_frames,
             "durations": durations,
@@ -229,6 +231,8 @@ def main():
     tone_id_map_path = dumpdir / "tone_id_map.txt"
     get_phones_tones(sentences, phone_id_map_path, tone_id_map_path,
                      args.dataset)
+    speaker_id_map_path = dumpdir / "speaker_id_map.txt"
+    get_spk_id_map(speaker_set, speaker_id_map_path)
 
     if args.dataset == "baker":
         wav_files = sorted(list((rootdir / "Wave").rglob("*.wav")))
@@ -239,6 +243,28 @@ def main():
         dev_wav_files = wav_files[num_train:num_train + num_dev]
         test_wav_files = wav_files[num_train + num_dev:]
 
+    elif args.dataset == "other":
+        sub_num_dev = 100
+        wav_dir = rootdir / "wav"
+        train_wav_files = []
+        dev_wav_files = []
+        test_wav_files = []
+        for speaker in os.listdir(wav_dir):
+            if os.path.exists(os.path.join(wav_dir, speaker, "split")):
+                wav_files = sorted(list((wav_dir / speaker / "split").rglob("*.wav")))
+            else:
+                wav_files = sorted(list((wav_dir / speaker).rglob("*.wav")))
+
+            if len(wav_files) > 100:
+                train_wav_files += wav_files[:-sub_num_dev * 2]
+                dev_wav_files += wav_files[-sub_num_dev * 2:-sub_num_dev]
+                test_wav_files += wav_files[-sub_num_dev:]
+            else:
+                train_wav_files += wav_files
+        print("len train_wav_files", len(train_wav_files))
+        print("len dev_wav_files", len(dev_wav_files))
+        print("len test_wav_files", len(test_wav_files))
+
     train_dump_dir = dumpdir / "train" / "raw"
     train_dump_dir.mkdir(parents=True, exist_ok=True)
     dev_dump_dir = dumpdir / "dev" / "raw"

paddlespeech/t2s/exps/speedyspeech/train.py

@@ -27,7 +27,8 @@ from paddle.io import DataLoader
 from paddle.io import DistributedBatchSampler
 from yacs.config import CfgNode
 
-from paddlespeech.t2s.datasets.am_batch_fn import speedyspeech_batch_fn
+from paddlespeech.t2s.datasets.am_batch_fn import speedyspeech_single_spk_batch_fn
+from paddlespeech.t2s.datasets.am_batch_fn import speedyspeech_multi_spk_batch_fn
 from paddlespeech.t2s.datasets.data_table import DataTable
 from paddlespeech.t2s.models.speedyspeech import SpeedySpeech
 from paddlespeech.t2s.models.speedyspeech import SpeedySpeechEvaluator
@@ -57,6 +58,21 @@ def train_sp(args, config):
         f"rank: {dist.get_rank()}, pid: {os.getpid()}, parent_pid: {os.getppid()}",
     )
 
+    fields = ["phones", "tones", "num_phones", "num_frames", "feats", "durations"]
+
+    spk_num = None
+    if args.speaker_dict is not None:
+        print("multiple speaker speedyspeech!")
+        collate_fn = speedyspeech_multi_spk_batch_fn
+        with open(args.speaker_dict, 'rt') as f:
+            spk_id = [line.strip().split() for line in f.readlines()]
+        spk_num = len(spk_id)
+        fields += ["spk_id"]
+    else:
+        print("single speaker speedyspeech!")
+        collate_fn = speedyspeech_single_spk_batch_fn
+    print("spk_num:", spk_num)
+
     # dataloader has been too verbose
     logging.getLogger("DataLoader").disabled = True
 
@@ -71,9 +87,7 @@ def train_sp(args, config):
 
     train_dataset = DataTable(
         data=train_metadata,
-        fields=[
-            "phones", "tones", "num_phones", "num_frames", "feats", "durations"
-        ],
+        fields=fields,
         converters={
             "feats": np.load,
         }, )
@@ -87,9 +101,7 @@ def train_sp(args, config):
 
     dev_dataset = DataTable(
         data=dev_metadata,
-        fields=[
-            "phones", "tones", "num_phones", "num_frames", "feats", "durations"
-        ],
+        fields=fields,
         converters={
             "feats": np.load,
         }, )
@@ -105,14 +117,14 @@ def train_sp(args, config):
     train_dataloader = DataLoader(
         train_dataset,
         batch_sampler=train_sampler,
-        collate_fn=speedyspeech_batch_fn,
+        collate_fn=collate_fn,
         num_workers=config.num_workers)
     dev_dataloader = DataLoader(
         dev_dataset,
         shuffle=False,
         drop_last=False,
         batch_size=config.batch_size,
-        collate_fn=speedyspeech_batch_fn,
+        collate_fn=collate_fn,
         num_workers=config.num_workers)
     print("dataloaders done!")
     with open(args.phones_dict, "r") as f:
@@ -125,7 +137,7 @@ def train_sp(args, config):
     print("tone_size:", tone_size)
 
     model = SpeedySpeech(
-        vocab_size=vocab_size, tone_size=tone_size, **config["model"])
+        vocab_size=vocab_size, tone_size=tone_size, spk_num=spk_num, **config["model"])
     if world_size > 1:
         model = DataParallel(model)
     print("model done!")
@@ -184,6 +196,12 @@ def main():
     parser.add_argument(
         "--tones-dict", type=str, default=None, help="tone vocabulary file.")
 
+    parser.add_argument(
+        "--speaker-dict",
+        type=str,
+        default=None,
+        help="speaker id map file for multiple speaker model.")
+
     # 这里可以多传入 max_epoch 等
     args, rest = parser.parse_known_args()
 

paddlespeech/t2s/models/speedyspeech/speedyspeech.py

@@ -14,7 +14,7 @@
 import numpy as np
 import paddle
 from paddle import nn
-
+import paddle.nn.functional as F
 from paddlespeech.t2s.modules.positional_encoding import sinusoid_position_encoding
 
 
@@ -171,7 +171,11 @@ class SpeedySpeech(nn.Layer):
             decoder_output_size,
             decoder_kernel_size,
             decoder_dilations,
-            tone_size=None, ):
+            tone_size=None, 
+            spk_num: int=None,
+            spk_embed_dim: int=None,
+            spk_embed_integration_type: str="add",
+            ):
         super().__init__()
         encoder = SpeedySpeechEncoder(vocab_size, tone_size,
                                       encoder_hidden_size, encoder_kernel_size,
@@ -183,14 +187,43 @@ class SpeedySpeech(nn.Layer):
         self.encoder = encoder
         self.duration_predictor = duration_predictor
         self.decoder = decoder
-
-    def forward(self, text, tones, durations):
+        self.spk_embed_dim = spk_embed_dim
+        # use idx 0 as padding idx
+        self.padding_idx = 0
+
+        if self.spk_embed_dim is not None:
+            self.spk_embed_integration_type = spk_embed_integration_type
+        if spk_num and self.spk_embed_dim:
+            self.spk_embedding_table = nn.Embedding(
+                num_embeddings=spk_num,
+                embedding_dim=self.spk_embed_dim,
+                padding_idx=self.padding_idx)
+        self.encoder_hidden_size = encoder_hidden_size
+        # define additional projection for speaker embedding
+        
+        if self.spk_embed_dim is not None:
+            print("spk_embed_integration_type------------", spk_embed_integration_type)
+            if self.spk_embed_integration_type == "add":
+                self.spk_projection = nn.Linear(self.spk_embed_dim, self.encoder_hidden_size)
+            else:
+                self.spk_projection = nn.Linear(self.encoder_hidden_size + self.spk_embed_dim, self.encoder_hidden_size)
+
+
+    def forward(self, text, tones, durations, spk_id: paddle.Tensor=None):
         # input of embedding must be int64
         text = paddle.cast(text, 'int64')
         tones = paddle.cast(tones, 'int64')
+        if spk_id is not None:
+            spk_id = paddle.cast(spk_id, 'int64')
         durations = paddle.cast(durations, 'int64')
         encodings = self.encoder(text, tones)
         # (B, T)
+
+        if self.spk_embed_dim is not None:
+            if spk_id is not None:
+                spk_emb = self.spk_embedding_table(spk_id)
+                encodings = self._integrate_with_spk_embed(encodings, spk_emb)
+
         pred_durations = self.duration_predictor(encodings.detach())
 
         # expand encodings
@@ -204,7 +237,7 @@ class SpeedySpeech(nn.Layer):
         decoded = self.decoder(encodings)
         return decoded, pred_durations
 
-    def inference(self, text, tones=None):
+    def inference(self, text, tones=None, spk_id=None,):
         # text: [T]
         # tones: [T]
         # input of embedding must be int64
@@ -215,6 +248,11 @@ class SpeedySpeech(nn.Layer):
             tones = tones.unsqueeze(0)
 
         encodings = self.encoder(text, tones)
+        if self.spk_embed_dim is not None:
+            if spk_id is not None:
+                spk_emb = self.spk_embedding_table(spk_id)
+                encodings = self._integrate_with_spk_embed(encodings, spk_emb)
+
         pred_durations = self.duration_predictor(encodings)  # (1, T)
         durations_to_expand = paddle.round(pred_durations.exp())
         durations_to_expand = (durations_to_expand).astype(paddle.int64)
@@ -240,6 +278,34 @@ class SpeedySpeech(nn.Layer):
         decoded = self.decoder(encodings)
         return decoded[0]
 
+    def _integrate_with_spk_embed(self, hs, spk_emb):
+        """Integrate speaker embedding with hidden states.
+
+        Parameters
+        ----------
+        hs : Tensor
+            Batch of hidden state sequences (B, Tmax, adim).
+        spk_emb : Tensor
+            Batch of speaker embeddings (B, spk_embed_dim).
+
+        Returns
+        ----------
+        Tensor
+            Batch of integrated hidden state sequences (B, Tmax, adim)
+        """
+        if self.spk_embed_integration_type == "add":
+            # apply projection and then add to hidden states
+            spk_emb = self.spk_projection(F.normalize(spk_emb))
+            hs = hs + spk_emb.unsqueeze(1)
+        elif self.spk_embed_integration_type == "concat":
+            # concat hidden states with spk embeds and then apply projection
+            spk_emb = F.normalize(spk_emb).unsqueeze(1).expand(
+                shape=[-1, hs.shape[1], -1])
+            hs = self.spk_projection(paddle.concat([hs, spk_emb], axis=-1))
+        else:
+            raise NotImplementedError("support only add or concat.")
+
+        return hs
 
 class SpeedySpeechInference(nn.Layer):
     def __init__(self, normalizer, speedyspeech_model):
@@ -247,7 +313,7 @@ class SpeedySpeechInference(nn.Layer):
         self.normalizer = normalizer
         self.acoustic_model = speedyspeech_model
 
-    def forward(self, phones, tones):
-        normalized_mel = self.acoustic_model.inference(phones, tones)
+    def forward(self, phones, tones, spk_id=None):
+        normalized_mel = self.acoustic_model.inference(phones, tones, spk_id)
         logmel = self.normalizer.inverse(normalized_mel)
         return logmel

paddlespeech/t2s/models/speedyspeech/speedyspeech_updater.py

@@ -50,10 +50,15 @@ class SpeedySpeechUpdater(StandardUpdater):
         self.msg = "Rank: {}, ".format(dist.get_rank())
         losses_dict = {}
 
+        # spk_id!=None in multiple spk speedyspeech 
+        spk_id = batch["spk_id"] if "spk_id" in batch else None
+
         decoded, predicted_durations = self.model(
             text=batch["phones"],
             tones=batch["tones"],
-            durations=batch["durations"])
+            durations=batch["durations"],
+            spk_id=spk_id
+            )
 
         target_mel = batch["feats"]
         spec_mask = F.sequence_mask(
@@ -112,10 +117,14 @@ class SpeedySpeechEvaluator(StandardEvaluator):
         self.msg = "Evaluate: "
         losses_dict = {}
 
+        spk_id = batch["spk_id"] if "spk_id" in batch else None
+
         decoded, predicted_durations = self.model(
             text=batch["phones"],
             tones=batch["tones"],
-            durations=batch["durations"])
+            durations=batch["durations"],
+            spk_id=spk_id
+            )
 
         target_mel = batch["feats"]
         spec_mask = F.sequence_mask(