rixty/KoboldAI-Client
1
1
Fork 0
mirror of https://github.com/KoboldAI/KoboldAI-Client.git synced 2025-06-05 21:59:24 +02:00
Code Issues Projects Releases Wiki Activity

Added functional audio generation (requires ffmpeg installed or ffmpeg.exe in koboldai directory)

Browse Source 
added load v1 story conversion to saved v2 story if loaded from disk
This commit is contained in:
ebolam
2022-12-05 20:52:38 -05:00
parent f77f6f7fba
commit 1f62f154a5
6 changed files with 86 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
aiserver.py
View File

@@ -7251,18 +7251,20 @@ def loadRequest(loadpath, filename=None):
if(isinstance(loadpath, str)):
with open(loadpath, "r") as file:
js = json.load(file)
from_file=loadpath
if(filename is None):
filename = path.basename(loadpath)
else:
js = loadpath
if(filename is None):
filename = "untitled.json"
from_file=None
js['v1_loadpath'] = loadpath
js['v1_filename'] = filename
logger.debug("Loading JSON data took {}s".format(time.time()-start_time))
loadJSON(js)
loadJSON(js, from_file=from_file)
def loadJSON(json_text_or_dict):
def loadJSON(json_text_or_dict, from_file=None):
logger.debug("Loading JSON Story")
logger.debug("Called from {}".format(inspect.stack()[1].function))
start_time = time.time()
@@ -7275,13 +7277,13 @@ def loadJSON(json_text_or_dict):
if json_data['file_version'] == 2:
load_story_v2(json_data)
else:
load_story_v1(json_data)
load_story_v1(json_data, from_file=from_file)
else:
load_story_v1(json_data)
load_story_v1(json_data, from_file=from_file)
logger.debug("Calcing AI Text from Story Load")
ignore = koboldai_vars.calc_ai_text()
def load_story_v1(js):
def load_story_v1(js, from_file=None):
logger.debug("Loading V1 Story")
logger.debug("Called from {}".format(inspect.stack()[1].function))
loadpath = js['v1_loadpath'] if 'v1_loadpath' in js else koboldai_vars.savedir
@@ -7402,6 +7404,12 @@ def load_story_v1(js):
print("{0}Story loaded from {1}!{2}".format(colors.GREEN, filename, colors.END))
send_debug()
if from_file is not None:
#Save the file so we get a new V2 format, then move the save file into the proper directory
koboldai_vars.save_story()
shutil.move(from_file, koboldai_vars.save_paths.story.replace("story.json", "v1_file.json"))
def load_story_v2(js):
logger.debug("Loading V2 Story")
@@ -9881,7 +9889,7 @@ def UI_2_test_match():
@app.route("/audio")
@logger.catch
def UI_2_audio():
action_id = int(request.args['id']) if 'id' in request.args else len(koboldai_vars.actions)
action_id = int(request.args['id']) if 'id' in request.args else koboldai_vars.actions.action_count
filename = os.path.join(koboldai_vars.save_paths.generated_audio, f"{action_id}.ogg")
if not os.path.exists(filename):
koboldai_vars.actions.gen_audio(action_id)

8
koboldai_settings.py
View File

@@ -1058,6 +1058,7 @@ class story_settings(settings):
self.assign_world_info_to_actions(action_id=-1, wuid=None)
process_variable_changes(self.socketio, self.__class__.__name__.replace("_settings", ""), 'prompt_wi_highlighted_text', self.prompt_wi_highlighted_text, None)
ignore = self.koboldai_vars.calc_ai_text()
self.actions.gen_audio(action_id=-1)
#Because we have seperate variables for action types, this syncs them
elif name == 'storymode':
@@ -1871,7 +1872,10 @@ class KoboldStoryRegister(object):
filename = os.path.join(self.koboldai_vars.save_paths.generated_audio, f"{action_id}.ogg")
if overwrite or not os.path.exists(filename):
self.make_audio_queue.put((self.actions[action_id]['Selected Text'], filename))
if action_id == -1:
self.make_audio_queue.put((self.koboldai_vars.prompt, filename))
else:
self.make_audio_queue.put((self.actions[action_id]['Selected Text'], filename))
if self.make_audio_thread is None or not self.make_audio_thread.is_alive():
self.make_audio_thread = threading.Thread(target=self.create_wave, args=(self.tts_model, self.make_audio_queue))
self.make_audio_thread.start()
@@ -1894,7 +1898,7 @@ class KoboldStoryRegister(object):
def gen_all_audio(self, overwrite=False):
if self.story_settings.gen_audio and self.koboldai_vars.experimental_features:
for i in reversed(list(self.actions.keys())):
for i in reversed(list(self.actions.keys())+[-1]):
self.gen_audio(i, overwrite=False)
else:
print("{} and {}".format(self.story_settings.gen_audio, self.koboldai_vars.experimental_features))

29
static/koboldai.css
View File

@@ -1386,20 +1386,20 @@ body {
grid-area: inputrow;
z-index: 2;
display: grid;
grid-template-areas: "textarea statusbar statusbar statusbar"
"textarea submit submit submit"
"textarea back redo retry";
grid-template-columns: auto 30px 30px 30px;
grid-template-areas: "textarea tts_controls statusbar statusbar statusbar"
"textarea tts_controls submit submit submit"
"textarea tts_controls back redo retry";
grid-template-columns: auto min-content 30px 30px 30px;
grid-template-rows: 0px 60px 40px;
gap: 1px;
}
.inputrow[story_storymode="1"] {
grid-template-areas: "adventure_mode textarea statusbar statusbar statusbar"
"adventure_mode textarea submit submit submit"
"adventure_mode textarea back redo retry";
grid-template-columns: 80px auto 30px 30px 30px;
grid-template-areas: "adventure_mode textarea tts_controls statusbar statusbar statusbar"
"adventure_mode textarea tts_controls submit submit submit"
"adventure_mode textarea tts_controls back redo retry";
grid-template-columns: 80px auto min-content 30px 30px 30px;
grid-template-rows: 0px 60px 40px;
gap: 1px;
@@ -1540,6 +1540,19 @@ body {
z-index: 9999;
}
.tts_controls.hidden[story_gen_audio="true"] {
display: inherit !important;
}
.inputrow .tts_controls {
grid-area: tts_controls;
padding: 0px;
height: 100%;
width: 100%;
text-align: center;
overflow: hidden;
}
.inputrow .back {
grid-area: back;
padding: 0px;

37
static/koboldai.js
View File

@@ -101,6 +101,8 @@ var finder_actions = [
const context_menu_actions = {
gamescreen: [
{label: "Speak", icon: "record_voice_over", enabledOn: "SELECTION", click: speak_audio},
null,
{label: "Cut", icon: "content_cut", enabledOn: "SELECTION", click: cut},
{label: "Copy", icon: "content_copy", enabledOn: "SELECTION", click: copy},
{label: "Paste", icon: "content_paste", enabledOn: "SELECTION", click: paste},
@@ -3165,6 +3167,41 @@ function retry_from_here() {
}
}
function speak_audio() {
let chunk = null;
for (element of document.getElementsByClassName("editing")) {
if (element.id == 'story_prompt') {
chunk = -1
} else {
chunk = parseInt(element.id.split(" ").at(-1));
}
element.classList.remove("editing");
}
if (chunk != null) {
action_count = parseInt(document.getElementById("action_count").textContent);
//console.log(chunk);
document.getElementById("reader").src = "/audio?id="+chunk;
document.getElementById("reader").play();
document.getElementById("play_tts").textContent = "pause";
}
}
function play_pause_tts() {
if (document.getElementById("reader").paused) {
document.getElementById("reader").play();
document.getElementById("play_tts").textContent = "pause";
} else {
document.getElementById("reader").pause();
document.getElementById("play_tts").textContent = "play_arrow";
}
}
function stop_tts() {
document.getElementById("reader").src="";
document.getElementById("reader").src="/audio";
document.getElementById("play_tts").textContent = "play_arrow";
}
function view_selection_probabilities() {
// Not quite sure how this should work yet. Probabilities are obviously on
// the token level, which we have no UI representation of. There are other

10
templates/index_new.html
View File

@@ -98,13 +98,17 @@
<div><span class="var_sync_model_horde_wait_time"></span> sec left</div>
</div>
</div><br>
<span class="tts_controls hidden var_sync_alt_story_gen_audio">
<button type="button" class="btn action_button" style="width: 30px; padding: 0px;" onclick='play_pause_tts()' aria-label="play"><span id="play_tts" class="material-icons-outlined" style="font-size: 1.4em;">play_arrow</span></button>
<button type="button" class="btn action_button" style="width: 30px; padding: 0px;" onclick='stop_tts()' aria-label="play"><span id="stop_tts" class="material-icons-outlined" style="font-size: 1.4em;">stop</span></button>
</span>
<button type="button" class="btn action_button submit var_sync_alt_system_aibusy" system_aibusy=False id="btnsubmit"
onclick="socket.emit('submit', {'data': document.getElementById('input_text').value, 'theme': document.getElementById('themetext').value});document.getElementById('input_text').value = '';document.getElementById('themetext').value = '';"
>Submit</button>
<button type="button" class="btn action_button submited var_sync_alt_system_aibusy" system_aibusy=False id="btnsent"><img id="thinking" src="static/thinking.gif" class="force_center" onclick="socket.emit('abort','');"></button>
<button type="button" class="btn action_button back" onclick="socket.emit('back', {});" aria-label="undo"><span class="oi" data-glyph="action-undo"></span></button>
<button type="button" class="btn action_button redo" onclick="socket.emit('redo', {});" aria-label="redo"><span class="oi" data-glyph="action-redo"></span></button>
<button type="button" class="btn action_button retry" onclick="socket.emit('retry', {});" aria-label="retry"><span class="oi" data-glyph="loop-circular"></span></button>
<button type="button" class="btn action_button back" onclick="socket.emit('back', {});" aria-label="undo"><span class="material-icons-outlined" style="font-size: 1.4em;">replay</span></button>
<button type="button" class="btn action_button redo" onclick="socket.emit('redo', {});" aria-label="redo"><span class="material-icons-outlined" style="font-size: 1.4em;">arrow_forward</span></button>
<button type="button" class="btn action_button retry" onclick="socket.emit('retry', {});" aria-label="retry"><span class="material-icons-outlined" style="font-size: 1.4em;">autorenew</span></button>
</div>
</div>

2
templates/settings flyout.html
View File

@@ -84,7 +84,7 @@
<hr/>
</div>
<div id="tts" class="var_sync_alt_system_experimental_features">
<audio controls id="reader" preload=none>
<audio id="reader" preload=none src="/audio" />
</div>
<div class="setting_tile_area">
{% with menu='Home' %}