jaidaken/ComfyUI
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

Add llm sampling options and make reference audio work on ace step 1.5 (#12295)

Browse Source
This commit is contained in:
comfyanonymous
2026-02-04 18:29:22 -08:00
committed by GitHub
parent c8fcbd66ee
commit 6125b80979
4 changed files with 47 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
comfy_extras/nodes_ace.py
View File

@@ -44,13 +44,18 @@ class TextEncodeAceStepAudio15(io.ComfyNode):
io.Combo.Input("timesignature", options=['2', '3', '4', '6']),
io.Combo.Input("language", options=["en", "ja", "zh", "es", "de", "fr", "pt", "ru", "it", "nl", "pl", "tr", "vi", "cs", "fa", "id", "ko", "uk", "hu", "ar", "sv", "ro", "el"]),
io.Combo.Input("keyscale", options=[f"{root} {quality}" for quality in ["major", "minor"] for root in ["C", "C#", "Db", "D", "D#", "Eb", "E", "F", "F#", "Gb", "G", "G#", "Ab", "A", "A#", "Bb", "B"]]),
io.Boolean.Input("generate_audio_codes", default=True, tooltip="Enable the LLM that generates audio codes. This can be slow but will increase the quality of the generated audio. Turn this off if you are giving the model an audio reference.", advanced=True),
io.Float.Input("cfg_scale", default=2.0, min=0.0, max=100.0, step=0.1, advanced=True),
io.Float.Input("temperature", default=0.85, min=0.0, max=2.0, step=0.01, advanced=True),
io.Float.Input("top_p", default=0.9, min=0.0, max=2000.0, step=0.01, advanced=True),
io.Int.Input("top_k", default=0, min=0, max=100, advanced=True),
],
outputs=[io.Conditioning.Output()],
)
@classmethod
def execute(cls, clip, tags, lyrics, seed, bpm, duration, timesignature, language, keyscale) -> io.NodeOutput:
tokens = clip.tokenize(tags, lyrics=lyrics, bpm=bpm, duration=duration, timesignature=int(timesignature), language=language, keyscale=keyscale, seed=seed)
def execute(cls, clip, tags, lyrics, seed, bpm, duration, timesignature, language, keyscale, generate_audio_codes, cfg_scale, temperature, top_p, top_k) -> io.NodeOutput:
tokens = clip.tokenize(tags, lyrics=lyrics, bpm=bpm, duration=duration, timesignature=int(timesignature), language=language, keyscale=keyscale, seed=seed, generate_audio_codes=generate_audio_codes, cfg_scale=cfg_scale, temperature=temperature, top_p=top_p, top_k=top_k)
conditioning = clip.encode_from_tokens_scheduled(tokens)
return io.NodeOutput(conditioning)
@@ -100,14 +105,15 @@ class EmptyAceStep15LatentAudio(io.ComfyNode):
latent = torch.zeros([batch_size, 64, length], device=comfy.model_management.intermediate_device())
return io.NodeOutput({"samples": latent, "type": "audio"})
class ReferenceTimbreAudio(io.ComfyNode):
class ReferenceAudio(io.ComfyNode):
@classmethod
def define_schema(cls):
return io.Schema(
node_id="ReferenceTimbreAudio",
display_name="Reference Audio",
category="advanced/conditioning/audio",
is_experimental=True,
description="This node sets the reference audio for timbre (for ace step 1.5)",
description="This node sets the reference audio for ace step 1.5",
inputs=[
io.Conditioning.Input("conditioning"),
io.Latent.Input("latent", optional=True),
@@ -131,7 +137,7 @@ class AceExtension(ComfyExtension):
EmptyAceStepLatentAudio,
TextEncodeAceStepAudio15,
EmptyAceStep15LatentAudio,
ReferenceTimbreAudio,
ReferenceAudio,
]
async def comfy_entrypoint() -> AceExtension: