Audiocraft is a library for audio processing and generation with deep learning. It features the state-of-the-art EnCodec audio compressor / tokenizer, along with MusicGen, a simple and controllable music generation LM with textual and melodic conditioning.
MIT License
20.15k
stars
2.01k
forks
source link
MusicGen: Missing 'rtf' assignment in prompted samples generation. #458
Hi, I'm currently experimenting on MusicGen and encountered a 'rtf undefined' exception when the system is configured to only generate prompted samples, that is the 'generate' section in config/solver/musigen/default.yaml is modified as:
generate:
every: 25
num_workers: 5
path: samples
audio:
format: wav
strategy: loudness
sample_rate: ${sample_rate}
loudness_headroom_db: 14
lm:
prompted_samples: true
unprompted_samples: false # <- this line is modified
gen_gt_samples: false
prompt_duration: null # if not set, will use dataset.generate.segment_duration / 4
gen_duration: null # if not set, will use dataset.generate.segment_duration
remove_prompts: false
# generation params
use_sampling: false
temp: 1.0
top_k: 0
top_p: 0.0
I'm guessing the assignment of 'rtf' is missing in the section for prompted sample generation?
Hi, I'm currently experimenting on MusicGen and encountered a 'rtf undefined' exception when the system is configured to only generate prompted samples, that is the 'generate' section in config/solver/musigen/default.yaml is modified as:
I'm guessing the assignment of 'rtf' is missing in the section for prompted sample generation?
Currently, I've modified the generation section for the 'rtf' metric as follow:
Please let me know if the modification is correct. Thanks for the great work.