SillyTavern/SillyTavern-extras
1
0
Fork 0
mirror of https://github.com/SillyTavern/SillyTavern-Extras.git synced 2026-04-24 08:28:59 +00:00
Code Issues Packages Projects Releases Wiki Activity

Added missing route to populate ST coqui UI settings and download/check models of coqui api.

Browse Source
This commit is contained in:
Tony Ribeiro
2023-08-13 02:19:47 +02:00
parent 1389d3806d
commit db63c58c30
2 changed files with 153 additions and 39 deletions
Download Patch File Download Diff File Expand all files Collapse all files

184
modules/text_to_speech/coqui/coqui_module.py
View File

@@ -17,6 +17,7 @@ import json
import os
import io
from TTS.api import TTS
from TTS.utils.manage import ModelManager
DEBUG_PREFIX = "<Coqui-TTS module>"
@@ -25,23 +26,19 @@ OUTPUT_PATH = "data/tmp/coqui_output.wav"
gpu = False
def coqui_supported_models():
def coqui_get_api_models():
"""
Return supported models in the following format: [language][dataset][name] = TTS_string_id
Example:
{
"multilingual": {
"multi-dataset": {
"your_tts": "tts_models/multilingual/multi-dataset/your_tts"
}
},
"en": {
"ljspeech": {
"tacotron2-DDC": "tts_models/en/ljspeech/tacotron2-DDC",
"glow-tts": "tts_models/en/ljspeech/glow-tts",
"vits": "tts_models/en/ljspeech/vits"
},
"jenny": {
"jenny": "tts_models/en/jenny/jenny"
"vctk": {
"vits": "tts_models/en/vctk/vits"
}
},
"ja": {
@@ -72,10 +69,13 @@ def coqui_supported_models():
if language not in models:
models[language] = {}
models[language][name+"/"+dataset] = i
response = json.dumps(models,indent=4)
if dataset not in models[language]:
models[language][dataset] = {}
models[language][dataset][name] = i
response = json.dumps(models)
return response
except Exception as e:
@@ -83,22 +83,137 @@ def coqui_supported_models():
abort(500, DEBUG_PREFIX + " Exception occurs while trying to get list of TTS available")
def coqui_check_model_state():
"""
Check if the requested model is installed on the server machine
"""
try:
model_state = "absent"
request_json = request.get_json()
model_id = request_json["model_id"]
print(DEBUG_PREFIX,"Search for model", model_id)
coqui_models_folder = ModelManager().output_prefix # models location
installed_models = os.listdir(coqui_models_folder)
model_folder_exists = False
model_folder = None
for i in installed_models:
if model_id == i.replace("--","/"):
model_folder_exists = True
model_folder = i
print(DEBUG_PREFIX,"Folder found:",model_folder)
# Check failed download
if model_folder_exists:
content = os.listdir(os.path.join(coqui_models_folder,model_folder))
print(DEBUG_PREFIX,"Checking content:",content)
for i in content:
if i == model_folder+".zip":
print("Corrupt installed found, model download must have failed previously")
model_state = "corrupted"
break
if model_state != "corrupted":
model_state = "installed"
response = json.dumps({"model_state":model_state})
return response
except Exception as e:
print(e)
abort(500, DEBUG_PREFIX + " Exception occurs while trying to search for installed model")
def coqui_install_model():
"""
Install requested model is installed on the server machine
"""
try:
model_installed = False
request_json = request.get_json()
model_id = request_json["model_id"]
print(DEBUG_PREFIX,"Search for model", model_id)
coqui_models_folder = ModelManager().output_prefix # models location
installed_models = os.listdir(coqui_models_folder)
for i in installed_models:
if model_id == i.replace("--","/"):
model_installed = True
response = json.dumps({"model_installed":model_installed})
return response
except Exception as e:
print(e)
abort(500, DEBUG_PREFIX + " Exception occurs while trying to search for installed model")
def coqui_get_local_models():
"""
Return user local models list in the following format: [language][dataset][name] = TTS_string_id
"""
abort(500, DEBUG_PREFIX + " Not implemented yet")
def coqui_get_model_settings():
"""
Process request model and return available speakers
- expected request: {
model_id: string
}
"""
try:
request_json = request.get_json()
#print(request_json)
print(DEBUG_PREFIX,"Received get_speakers request for model", request_json["model_id"])
model_id = request_json["model_id"]
model_languages = []
model_speakers = []
print(DEBUG_PREFIX,"Loading tts model", model_id,"\n - using", ("GPU" if gpu else "CPU"))
tts = TTS(model_name=model_id, progress_bar=True, gpu=gpu)
if tts.is_multi_lingual:
model_languages = tts.languages
if tts.is_multi_speaker:
model_speakers = tts.speakers
response = json.dumps({"languages":model_languages, "speakers":model_speakers})
print(DEBUG_PREFIX,"Model settings: ", response)
return response
except Exception as e:
print(e)
abort(500, DEBUG_PREFIX + " Exception occurs while trying to get model speakers")
def coqui_process_text():
"""
Process request text with the loaded RVC model
- expected request: {
text: string,
voiceId: string
}
- expected request: {
"text": text,
"model_id": voiceId,
"language": language,
"speaker": speaker
}
voiceId formats:
- model_type/language/dataset/model_name
- model_type/language/dataset/model_name[spearker_id]
- model_type/language/dataset/model_name[spearker_id][language_id]
examples:
- tts_models/ja/kokoro/tacotron2-DDC
- tts_models/en/vctk/vits[0]
- tts_models/multilingual/multi-dataset/your_tts[2][1]
- model_id formats:
- model_type/language/dataset/model_name
- model_type/language/dataset/model_name[spearker_id]
- model_type/language/dataset/model_name[spearker_id][language_id]
- examples:
- tts_models/ja/kokoro/tacotron2-DDC
- tts_models/en/vctk/vits[0]
- tts_models/multilingual/multi-dataset/your_tts[2][1]
"""
global gpu
@@ -106,21 +221,18 @@ def coqui_process_text():
request_json = request.get_json()
#print(request_json)
print(DEBUG_PREFIX,"Received TTS request for voiceId", request_json["voiceId"], "with text:\n",request_json["text"])
print(DEBUG_PREFIX,"Received TTS request for ", request_json)
text = request_json["text"]
tokens = [i.strip("]") for i in request_json["voiceId"].split("[")]
print(tokens)
model_name = tokens[0]
speaker = None
model_name = request_json["model_id"]
language = None
speaker = None
if len(tokens) > 1:
speaker = tokens[1]
if len(tokens) > 2:
language = tokens[2]
if request_json["language"] != "none":
language = request_json["language"]
if request_json["speaker"] != "none":
speaker = request_json["speaker"]
print(DEBUG_PREFIX,"Loading tts model", model_name, "\n - speaker: ",speaker,"\n - language: ",language, "\n - using",("GPU" if gpu else "CPU"))
@@ -130,14 +242,10 @@ def coqui_process_text():
if speaker is None:
abort(400, DEBUG_PREFIX + " Requested model "+model_name+" is multi-speaker but no speaker provided")
speaker = tts.speakers[int(speaker)]
if tts.is_multi_lingual:
if speaker is None:
abort(400, DEBUG_PREFIX + " Requested model "+model_name+" is multi-lingual but no language provided")
language = tts.languages[int(language)]
tts.tts_to_file(text=text, file_path=OUTPUT_PATH, speaker=speaker, language=language)
print(DEBUG_PREFIX, "Success, saved to",OUTPUT_PATH)

8
server.py
View File

@@ -385,7 +385,13 @@ if "coqui-tts" in modules:
import modules.text_to_speech.coqui.coqui_module as coqui_module
if mode == "GPU":
coqui_module.gpu = True
app.add_url_rule("/api/text-to-speech/coqui/supported-models", view_func=coqui_module.coqui_supported_models, methods=["POST"])
app.add_url_rule("/api/text-to-speech/coqui/coqui-api/get-models", view_func=coqui_module.coqui_get_api_models, methods=["POST"])
app.add_url_rule("/api/text-to-speech/coqui/coqui-api/check-model-state", view_func=coqui_module.coqui_check_model_state, methods=["POST"])
app.add_url_rule("/api/text-to-speech/coqui/coqui-api/install-model", view_func=coqui_module.coqui_install_model, methods=["POST"])
app.add_url_rule("/api/text-to-speech/coqui/coqui-api/get-model-settings", view_func=coqui_module.coqui_get_model_settings, methods=["POST"])
app.add_url_rule("/api/text-to-speech/coqui/local/get-models", view_func=coqui_module.coqui_get_local_models, methods=["POST"])
app.add_url_rule("/api/text-to-speech/coqui/process-text", view_func=coqui_module.coqui_process_text, methods=["POST"])
def require_module(name):