mrq/ai-voice-cloning
31
74
Fork 43
Code Issues 223 Pull Requests 5 Packages Projects Releases Wiki Activity

updated vall-e training template to use path-based speakers because it would just have a batch/epoch size of 1 otherwise; revert hardcoded 'spit processed dataset to this path' from my training rig to spit it out in a sane spot

Browse Source
This commit is contained in:
mrq 2023-08-24 21:45:50 +00:00
parent 533b73e083
commit a657623cbc
3 changed files with 8 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
models/.template.valle.yaml
View File

@ -23,7 +23,7 @@ dataset:
max_prompts: 3 max_prompts: 3
prompt_duration: 3.0 prompt_duration: 3.0
sample_type: speaker sample_type: path
tasks_list: ["tts"] # ["tts", "ns", "sr", "tse", "cse", "nse", "tts"] tasks_list: ["tts"] # ["tts", "ns", "sr", "tse", "cse", "nse", "tts"]

12
src/utils.py
View File

@ -2662,11 +2662,11 @@ def prepare_dataset( voice, use_segments=False, text_length=0, audio_length=0, p
if culled or args.tts_backend != "vall-e": if culled or args.tts_backend != "vall-e":
continue continue
# os.makedirs(f'{indir}/valle/', exist_ok=True) os.makedirs(f'{indir}/valle/', exist_ok=True)
os.makedirs(f'./training/valle/data/{voice}/', exist_ok=True) #os.makedirs(f'./training/valle/data/{voice}/', exist_ok=True)
#phn_file = f'{indir}/valle/{file.replace(f".{extension}",".phn.txt")}' phn_file = f'{indir}/valle/{file.replace(f".{extension}",".phn.txt")}'
phn_file = f'./training/valle/data/{voice}/{file.replace(f".{extension}",".phn.txt")}' #phn_file = f'./training/valle/data/{voice}/{file.replace(f".{extension}",".phn.txt")}'
if not os.path.exists(phn_file): if not os.path.exists(phn_file):
jobs['phonemize'][0].append(phn_file) jobs['phonemize'][0].append(phn_file)
jobs['phonemize'][1].append(normalized) jobs['phonemize'][1].append(normalized)
@ -2676,8 +2676,8 @@ def prepare_dataset( voice, use_segments=False, text_length=0, audio_length=0, p
print("Phonemized:", file, normalized, text) print("Phonemized:", file, normalized, text)
""" """
#qnt_file = f'{indir}/valle/{file.replace(f".{extension}",".qnt.pt")}' qnt_file = f'{indir}/valle/{file.replace(f".{extension}",".qnt.pt")}'
qnt_file = f'./training/valle/data/{voice}/{file.replace(f".{extension}",".qnt.pt")}' #qnt_file = f'./training/valle/data/{voice}/{file.replace(f".{extension}",".qnt.pt")}'
if 'error' not in result: if 'error' not in result:
if not quantize_in_memory and not os.path.exists(path): if not quantize_in_memory and not os.path.exists(path):
message = f"Missing segment, skipping... {file}" message = f"Missing segment, skipping... {file}"

2
src/webui.py
View File

@ -411,7 +411,7 @@ def setup_gradio():
GENERATE_SETTINGS["num_autoregressive_samples"] = gr.Slider(value=16, minimum=2, maximum=2048 if args.tts_backend=="vall-e" else 512, step=1, label="Samples", visible=args.tts_backend!="bark") GENERATE_SETTINGS["num_autoregressive_samples"] = gr.Slider(value=16, minimum=2, maximum=2048 if args.tts_backend=="vall-e" else 512, step=1, label="Samples", visible=args.tts_backend!="bark")
GENERATE_SETTINGS["diffusion_iterations"] = gr.Slider(value=30, minimum=0, maximum=512, step=1, label="Iterations", visible=args.tts_backend=="tortoise") GENERATE_SETTINGS["diffusion_iterations"] = gr.Slider(value=30, minimum=0, maximum=512, step=1, label="Iterations", visible=args.tts_backend=="tortoise")
GENERATE_SETTINGS["temperature"] = gr.Slider(value=0.95 if args.tts_backend=="vall-e" else 0.2, minimum=0, maximum=1, step=0.1, label="Temperature") GENERATE_SETTINGS["temperature"] = gr.Slider(value=0.95 if args.tts_backend=="vall-e" else 0.2, minimum=0, maximum=1, step=0.05, label="Temperature")
show_experimental_settings = gr.Checkbox(label="Show Experimental Settings", visible=args.tts_backend=="tortoise") show_experimental_settings = gr.Checkbox(label="Show Experimental Settings", visible=args.tts_backend=="tortoise")
reset_generate_settings_button = gr.Button(value="Reset to Default") reset_generate_settings_button = gr.Button(value="Reset to Default")