theluyuan/DiffSynth-Studio
1
0
Fork 0
mirror of https://github.com/modelscope/DiffSynth-Studio.git synced 2026-04-24 15:06:17 +00:00
Code Issues Packages Projects Releases Wiki Activity

path

Browse Source
This commit is contained in:
mi804
2026-04-23 18:09:16 +08:00
parent 002e3cdb74
commit 3da625432e
4 changed files with 8 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
examples/ace_step/model_inference/acestep-v15-base-CoverTask.py
View File

@@ -20,10 +20,10 @@ lyrics = '[Intro - Synth Brass Fanfare]\n\n[Verse 1]\n黑夜里的风吹过耳
dataset_snapshot_download( dataset_snapshot_download(
dataset_id="DiffSynth-Studio/diffsynth_example_dataset", dataset_id="DiffSynth-Studio/diffsynth_example_dataset",
local_dir="data/diffsynth_example_dataset", local_dir="data/diffsynth_example_dataset",
allow_file_pattern="ace_step/acestep-v15-base/audio.wav", allow_file_pattern="ace_step/acestep-v15-base-CoverTask/audio.wav",
) )
src_audio, sr = read_audio("data/diffsynth_example_dataset/ace_step/acestep-v15-base/audio.wav", resample=True, resample_rate=pipe.vae.sampling_rate) src_audio, sr = read_audio("data/diffsynth_example_dataset/ace_step/acestep-v15-base-CoverTask/audio.wav", resample=True, resample_rate=pipe.vae.sampling_rate)
# audio_cover_strength controls the steps of doing cover tasks. [0, num_inference_steps * audio_cover_strength] steps will be cover steps, and the rest will be regular text-to-music generation steps. # audio_cover_strength controls the steps of doing cover tasks. [0, num_inference_steps * audio_cover_strength] steps will be cover steps, and the rest will be regular text-to-music generation steps.
# denoising_strength controls how the output audio is influenced by the source audio in cover tasks. # denoising_strength controls how the output audio is influenced by the source audio in cover tasks.
audio = pipe( audio = pipe(

4
examples/ace_step/model_inference/acestep-v15-base-RepaintTask.py
View File

@@ -20,10 +20,10 @@ lyrics = '[Intro - Synth Brass Fanfare]\n\n[Verse 1]\n黑夜里的风吹过耳
dataset_snapshot_download( dataset_snapshot_download(
dataset_id="DiffSynth-Studio/diffsynth_example_dataset", dataset_id="DiffSynth-Studio/diffsynth_example_dataset",
local_dir="data/diffsynth_example_dataset", local_dir="data/diffsynth_example_dataset",
allow_file_pattern="ace_step/acestep-v15-base/audio.wav", allow_file_pattern="ace_step/acestep-v15-base-RepaintTask/audio.wav",
) )
src_audio, sr = read_audio("data/diffsynth_example_dataset/ace_step/acestep-v15-base/audio.wav", resample=True, resample_rate=pipe.vae.sampling_rate) src_audio, sr = read_audio("data/diffsynth_example_dataset/ace_step/acestep-v15-base-RepaintTask/audio.wav", resample=True, resample_rate=pipe.vae.sampling_rate)
# repainting_ranges are in seconds, and will be converted to frames internally in the pipeline. The negative value in repainting_ranges means the padding from the start of the audio. # repainting_ranges are in seconds, and will be converted to frames internally in the pipeline. The negative value in repainting_ranges means the padding from the start of the audio.
# For example, repainting_ranges=[(-10, 30), (160, 200)] means we want to repaint the audio from -10s to 30s (with 10s padding before the start) and from 160s to 200s. The non-existent parts will be padded with silence. # For example, repainting_ranges=[(-10, 30), (160, 200)] means we want to repaint the audio from -10s to 30s (with 10s padding before the start) and from 160s to 200s. The non-existent parts will be padded with silence.
# Repainting strength denotes the intensity of repainting area, where 0 means no repainting (keep the original audio) and 1 means full repainting. # Repainting strength denotes the intensity of repainting area, where 0 means no repainting (keep the original audio) and 1 means full repainting.

4
examples/ace_step/model_inference_low_vram/acestep-v15-base-CoverTask.py
View File

@@ -32,10 +32,10 @@ lyrics = '[Intro - Synth Brass Fanfare]\n\n[Verse 1]\n黑夜里的风吹过耳
dataset_snapshot_download( dataset_snapshot_download(
dataset_id="DiffSynth-Studio/diffsynth_example_dataset", dataset_id="DiffSynth-Studio/diffsynth_example_dataset",
local_dir="data/diffsynth_example_dataset", local_dir="data/diffsynth_example_dataset",
allow_file_pattern="ace_step/acestep-v15-base/audio.wav", allow_file_pattern="ace_step/acestep-v15-base-CoverTask/audio.wav",
) )
src_audio, sr = read_audio("data/diffsynth_example_dataset/ace_step/acestep-v15-base/audio.wav", resample=True, resample_rate=pipe.vae.sampling_rate) src_audio, sr = read_audio("data/diffsynth_example_dataset/ace_step/acestep-v15-base-CoverTask/audio.wav", resample=True, resample_rate=pipe.vae.sampling_rate)
# audio_cover_strength controls the steps of doing cover tasks. [0, num_inference_steps * audio_cover_strength] steps will be cover steps, and the rest will be regular text-to-music generation steps. # audio_cover_strength controls the steps of doing cover tasks. [0, num_inference_steps * audio_cover_strength] steps will be cover steps, and the rest will be regular text-to-music generation steps.
# denoising_strength controls how the output audio is influenced by the source audio in cover tasks. # denoising_strength controls how the output audio is influenced by the source audio in cover tasks.
audio = pipe( audio = pipe(

4
examples/ace_step/model_inference_low_vram/acestep-v15-base-RepaintTask.py
View File

@@ -32,10 +32,10 @@ lyrics = '[Intro - Synth Brass Fanfare]\n\n[Verse 1]\n黑夜里的风吹过耳
dataset_snapshot_download( dataset_snapshot_download(
dataset_id="DiffSynth-Studio/diffsynth_example_dataset", dataset_id="DiffSynth-Studio/diffsynth_example_dataset",
local_dir="data/diffsynth_example_dataset", local_dir="data/diffsynth_example_dataset",
allow_file_pattern="ace_step/acestep-v15-base/audio.wav", allow_file_pattern="ace_step/acestep-v15-base-RepaintTask/audio.wav",
) )
src_audio, sr = read_audio("data/diffsynth_example_dataset/ace_step/acestep-v15-base/audio.wav", resample=True, resample_rate=pipe.vae.sampling_rate) src_audio, sr = read_audio("data/diffsynth_example_dataset/ace_step/acestep-v15-base-RepaintTask/audio.wav", resample=True, resample_rate=pipe.vae.sampling_rate)
# repainting_ranges are in seconds, and will be converted to frames internally in the pipeline. The negative value in repainting_ranges means the padding from the start of the audio. # repainting_ranges are in seconds, and will be converted to frames internally in the pipeline. The negative value in repainting_ranges means the padding from the start of the audio.
# For example, repainting_ranges=[(-10, 30), (160, 200)] means we want to repaint the audio from -10s to 30s (with 10s padding before the start) and from 160s to 200s. The non-existent parts will be padded with silence. # For example, repainting_ranges=[(-10, 30), (160, 200)] means we want to repaint the audio from -10s to 30s (with 10s padding before the start) and from 160s to 200s. The non-existent parts will be padded with silence.
# Repainting strength denotes the intensity of repainting area, where 0 means no repainting (keep the original audio) and 1 means full repainting. # Repainting strength denotes the intensity of repainting area, where 0 means no repainting (keep the original audio) and 1 means full repainting.