mrq/vall-e
5
9
Fork 0
Code Issues 8 Pull Requests Packages Projects Releases Wiki Activity

setting up for allowing training for a partial amount of the speechx tasks (do NOT try this at home yet without a proper model, as performance is predecated on having a solid base vall-e model for the tasks

Browse Source
This commit is contained in:
mrq 2023-08-19 00:16:08 -05:00
parent ae9d38aa31
commit 8f42c578c9
3 changed files with 20 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
data/config.yaml
View File

@ -20,6 +20,7 @@ dataset:
prompt_duration: 3.0
sample_type: speaker
tasks_list: ["tts"] # do NOT change this until you're ready to train for SpeechX tasks # ["tts", "tts", "ns", "sr", "tse", "tts", "tts"]
models:
_models:
@ -102,6 +103,7 @@ trainer:
inference:
use_vocos: True
normalize: False # do NOT change this unless you know exactly what you are doing.
bitsandbytes:
enabled: false

5
vall_e/config.py
View File

@ -132,6 +132,8 @@ class Dataset:
prompt_duration: float = 3.0
sample_type: str = "path" # path | speaker
tasks_list: list[str] = field(default_factory=lambda: ["tts"])
@dataclass()
class Model:
@ -475,7 +477,8 @@ try:
if not cfg.dataset.use_hdf5:
cfg.dataset.training = [ Path(dir) for dir in cfg.dataset.training ]
cfg.dataset.validation = [ Path(dir) for dir in cfg.dataset.validation ]
cfg.dataset.noise = [ Path(dir) for dir in cfg.dataset.noise ]
cfg.dataset.noise = [ Path(dir) for dir in cfg.dataset.noise ]
except Exception as e:
pass

22
vall_e/data.py
View File

@ -204,7 +204,7 @@ class Dataset(_Dataset):
@cached_property
def tasks(self):
return ["tts"] # "ns", "sr", "tse", "cse", "nse"
return cfg.dataset.tasks_list # ["tts", "tts", "ns", "sr", "tse", "tts", "tts"] # , "cse", "nse"
def _get_phone_symmap(self):
return get_phone_symmap()
@ -216,16 +216,22 @@ class Dataset(_Dataset):
return get_task_symmap()
def get_task_token( self, token ):
if not hasattr(self, "task_symmap"):
self.task_symmap = self._get_task_symmap()
return torch.Tensor([[ self.task_symmap[f'<{token}>'] for _ in range(cfg.models.prom_levels) ]]).to(dtype=torch.int16)
def sample_noise(self):
def sample_noise(self):
paths = []
print(cfg.dataset.noise)
for data_dir in cfg.dataset.noise:
paths.extend(data_dir.rglob("*.qnt.pt"))
path = random.choice(paths)
return _load_quants(path)
if False and cfg.dataset.use_hdf5:
key = f'/noise/{_get_hdf5_path(path)}'
qnt = torch.from_numpy(cfg.hdf5[key]["audio"][:, :cfg.models.prom_levels]).to(torch.int16)
else:
qnt = _load_quants(path)
return qnt
def sample_speakers(self, ignore=[]):
choices = set(self.spkrs) - set(ignore)
@ -305,7 +311,7 @@ class Dataset(_Dataset):
# extend the noise to fill the target audio
noise = repeat_extend_audio(noise, resps.shape[0])
# create the input prompt by merging the target audio with the noise
proms = merge_audio(resps, noise, scale=[1, 0.125])
proms = merge_audio(resps, noise, scale=[1, 0.125], device="cpu")
# set the target to just be the noise if <sr>
if task == "sr":
resps = noise
@ -325,10 +331,10 @@ class Dataset(_Dataset):
smallest_size = min(resps.shape[0], other_proms.shape[0])
if other_proms.shape[0] == smallest_size:
noisy_proms = merge_audio( resps[:smallest_size, :], other_proms, scale=[1, random.uniform(0.5, 0.75)] )
noisy_proms = merge_audio( resps[:smallest_size, :], other_proms, scale=[1, random.uniform(0.5, 0.75)], device="cpu" )
noisy_proms = torch.cat( [ noisy_proms, resps[smallest_size:, :] ] )
else:
noisy_proms = merge_audio( resps, other_proms[:smallest_size, :], scale=[1, random.uniform(0.5, 0.75)] )
noisy_proms = merge_audio( resps, other_proms[:smallest_size, :], scale=[1, random.uniform(0.5, 0.75)], device="cpu" )
noisy_proms = torch.cat( [ noisy_proms, other_proms[smallest_size:, :] ] )
# stitch together the promps