Skip to content

Convert AceStep to use CPP instead of Python. #4

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
kalebbroo wants to merge 4 commits into
base: main
Choose a base branch
from
Open

Convert AceStep to use CPP instead of Python. #4

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
218b24a
Convert AceStep to use CPP instead of Python.
kalebbroo Apr 18, 2026
7dd9626
fix core params
kalebbroo Apr 18, 2026
0faff06
setps and cfg should use core params
kalebbroo Apr 18, 2026
1baddb1
fix download location and model locations
kalebbroo Apr 19, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 4 additions & 2 deletions Assets/audio-integration.js
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,7 +31,7 @@ const AudioLabConfig = {
moonshine_stt: { category: 'audiolab_stt', providerFlag: 'moonshine_stt_params' },
realtimestt_stt: { category: 'audiolab_stt', providerFlag: 'realtimestt_params' },
musicgen_music: { category: 'audiolab_audiogen', providerFlag: 'musicgen_music_params', extraFlags: ['audiocraft_sampling'] },
acestep_music: { category: 'audiolab_audiogen', providerFlag: 'acestep_music_params', extraFlags: ['acestep_lm_params', 'acestep_task_params'] },
acestep_music: { category: 'audiolab_audiogen', providerFlag: 'acestep_music_params', extraFlags: ['acestep_lm_params', 'acestep_task_params'], keepCoreParams: ['steps', 'cfgscale'] },
openvoice_clone: { category: 'audiolab_clone', providerFlag: 'openvoice_clone_params' },
rvc_clone: { category: 'audiolab_clone', providerFlag: 'rvc_clone_params' },
gptsovits_clone: { category: 'audiolab_clone', providerFlag: 'gptsovits_clone_params' },
Expand Down Expand Up @@ -141,10 +141,12 @@ featureSetChangers.push(() => {

const curArch = currentModelHelper.curArch;
const isAudioModel = AudioLabConfig.isAudioModel(curArch);
const archConfig = isAudioModel ? AudioLabConfig.archToCategory[curArch] : null;
const keepCoreParams = archConfig?.keepCoreParams || [];
console.log('[audiolab] featureSetChanger: curArch =', JSON.stringify(curArch), 'isAudioModel =', isAudioModel);

for (const param of gen_param_types) {
if (AudioLabConfig.coreParamsToHide.includes(param.id)) {
if (AudioLabConfig.coreParamsToHide.includes(param.id) && !keepCoreParams.includes(param.id)) {
if (isAudioModel) {
if (!param.hasOwnProperty('original_feature_flag_audiolab')) {
param.original_feature_flag_audiolab = param.feature_flag;
Expand Down
58 changes: 28 additions & 30 deletions AudioBackends/DynamicAudioBackend.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1097,48 +1097,46 @@ private static Dictionary<string, object> BuildEngineArgs(T2IParamInput input, A
break;

case "acestep_music":
// Core DiT params (acestep_music_params)
// Resolve DiT GGUF filename from model config
string ditModel = modelDef?.EngineConfig?.TryGetValue("dit_model", out object dmObj) == true ? dmObj?.ToString() : "acestep-v15-turbo";
args["dit_file"] = AceStepCppManager.GetDitFileName(ditModel);
// AceRequest-compatible params
args["caption"] = input.Get(T2IParamTypes.Prompt, "");
args["lyrics"] = input.TryGet(AudioLabParams.Lyrics, out string ly) ? ly : "[Instrumental]";
args["seed"] = input.TryGet(T2IParamTypes.Seed, out long aceSeed) ? aceSeed : -1L;
args["infer_step"] = input.TryGet(AudioLabParams.InferStep, out int infStep) ? infStep : 8;
args["guidance_scale"] = input.TryGet(AudioLabParams.ACEGuidanceScale, out double aceGuide) ? aceGuide : 7.0;
args["instrumental"] = input.TryGet(AudioLabParams.Instrumental, out string aceInst) ? aceInst : "false";
args["bpm"] = input.TryGet(AudioLabParams.BPM, out int aceBpm) ? aceBpm : 120;
if (input.TryGet(AudioLabParams.KeyScale, out string aceKey) && !string.IsNullOrEmpty(aceKey))
args["key_scale"] = aceKey;
args["time_signature"] = input.TryGet(AudioLabParams.TimeSignature, out string aceTs) ? aceTs : "4";
args["vocal_language"] = input.TryGet(AudioLabParams.VocalLanguage, out string aceVl) ? aceVl : "en";
args["shift"] = input.TryGet(AudioLabParams.ACEShift, out double aceShift) ? aceShift : 3.0;
args["infer_method"] = input.TryGet(AudioLabParams.InferMethod, out string aceIm) ? aceIm : "ode";
args["use_adg"] = input.TryGet(AudioLabParams.UseADG, out string aceAdg) ? aceAdg : "false";
args["cfg_interval_start"] = input.TryGet(AudioLabParams.CFGIntervalStart, out double aceCfgS) ? aceCfgS : 0.0;
args["cfg_interval_end"] = input.TryGet(AudioLabParams.CFGIntervalEnd, out double aceCfgE) ? aceCfgE : 1.0;
args["enable_normalization"] = input.TryGet(AudioLabParams.EnableNormalization, out string aceNorm) ? aceNorm : "true";
args["normalization_db"] = input.TryGet(AudioLabParams.NormalizationDB, out double aceNormDb) ? aceNormDb : -14.0;
// LM planner params (acestep_lm_params) — TODO: integrate with SwarmUI AbstractLLMBackend
args["duration"] = input.TryGet(T2IParamTypes.Text2AudioDuration, out double aceDur) ? aceDur : 30.0;
args["inference_steps"] = input.TryGet(T2IParamTypes.Steps, out int infStep) ? infStep : 0;
args["guidance_scale"] = input.TryGet(T2IParamTypes.CFGScale, out double aceGuide) ? aceGuide : 0.0;
args["bpm"] = input.TryGet(T2IParamTypes.Text2AudioBPM, out long aceBpm) ? (int)aceBpm : 0;
args["keyscale"] = input.TryGet(T2IParamTypes.Text2AudioKeyScale, out string aceKey) ? aceKey : "";
args["timesignature"] = input.TryGet(T2IParamTypes.Text2AudioTimeSignature, out string aceTs) ? aceTs : "";
args["vocal_language"] = input.TryGet(T2IParamTypes.Text2AudioLanguage, out string aceVl) ? aceVl : "";
args["shift"] = input.TryGet(AudioLabParams.ACEShift, out double aceShift) ? aceShift : 0.0;
// LM planner params (acestep_lm_params)
args["lm_model"] = input.TryGet(AudioLabParams.ACELMModel, out string aceLm) ? aceLm : "none";
args["thinking"] = input.TryGet(AudioLabParams.Thinking, out string aceThink) ? aceThink : "true";
args["lm_temperature"] = input.TryGet(AudioLabParams.LMTemperature, out double aceLmTemp) ? aceLmTemp : 0.85;
args["lm_cfg_scale"] = input.TryGet(AudioLabParams.LMCFGScale, out double aceLmCfg) ? aceLmCfg : 2.0;
args["lm_top_k"] = input.TryGet(AudioLabParams.LMTopK, out int aceLmTopK) ? aceLmTopK : 0;
args["lm_top_p"] = input.TryGet(AudioLabParams.LMTopP, out double aceLmTopP) ? aceLmTopP : 0.9;
if (input.TryGet(AudioLabParams.LMNegativePrompt, out string aceLmNeg) && !string.IsNullOrEmpty(aceLmNeg))
args["lm_negative_prompt"] = aceLmNeg;
args["use_cot_metas"] = input.TryGet(AudioLabParams.UseCotMetas, out string aceCotM) ? aceCotM : "true";
args["use_cot_caption"] = input.TryGet(AudioLabParams.UseCotCaption, out string aceCotC) ? aceCotC : "true";
args["use_cot_language"] = input.TryGet(AudioLabParams.UseCotLanguage, out string aceCotL) ? aceCotL : "true";
// Task params (acestep_task_params)
args["task_type"] = input.TryGet(AudioLabParams.ACETaskType, out string aceTask) ? aceTask : "text2music";
args["use_cot_caption"] = input.TryGet(AudioLabParams.UseCotCaption, out string aceCotC) && aceCotC == "true";
// Task params — source audio determines task mode (acestep_task_params)
string aceSrcAudio = GetBase64Audio(input, AudioLabParams.ACESourceAudio);
if (!string.IsNullOrEmpty(aceSrcAudio))
{
args["src_audio"] = aceSrcAudio;
string aceRefAudio = GetBase64Audio(input, AudioLabParams.ACEReferenceAudio);
if (!string.IsNullOrEmpty(aceRefAudio))
args["reference_audio"] = aceRefAudio;
args["repaint_start"] = input.TryGet(AudioLabParams.RepaintStart, out double aceRpS) ? aceRpS : 0.0;
args["repaint_end"] = input.TryGet(AudioLabParams.RepaintEnd, out double aceRpE) ? aceRpE : -1.0;
args["cover_strength"] = input.TryGet(AudioLabParams.CoverStrength, out double aceCovStr) ? aceCovStr : 1.0;
args["cover_noise_strength"] = input.TryGet(AudioLabParams.CoverNoiseStrength, out double aceCovNs) ? aceCovNs : 0.0;
string taskType = input.TryGet(AudioLabParams.ACETaskType, out string aceTask) ? aceTask : "cover";
if (taskType == "cover")
args["audio_cover_strength"] = input.TryGet(AudioLabParams.CoverStrength, out double aceCovStr) ? aceCovStr : 0.5;
if (taskType == "repaint")
{
args["repainting_start"] = input.TryGet(AudioLabParams.RepaintStart, out double aceRpS) ? aceRpS : -1.0;
args["repainting_end"] = input.TryGet(AudioLabParams.RepaintEnd, out double aceRpE) ? aceRpE : -1.0;
}
if (taskType == "lego")
args["lego"] = "vocals";
}
break;

case "musicgen_music":
Expand Down
Loading