search

breizhn / DTLN

Tensorflow 2.x implementation of the DTLN real time speech denoising model. With TF-lite, ONNX and real-time audio processing support.
MIT License
567 stars 160 forks source link

Validation loss is too high #65

Open jeungmin717 opened 1 year ago

jeungmin717 commented 1 year ago

Thanks Breizhn for sharing this amazing repository.

I followed step by step reading documentation. By using train data from you provided ( breizhn/DNS-challenge repository) I get validation loss around 49.29 training loss around 0.0014 epoch_lr : decreases from 0.0010 to 1.5259e-08

The validation loss is over too high. I don't think the data is corrupt, since I am using the same training_data repository you provided. Can I get some hint or help? How can I get correct validation loss .

Thanks in advance.

noisyspeech_synthesizer.cfg I used. didn't change a thing

sampling_rate: 16000 audioformat: *.wav audio_length: 30 silence_length: 0.2 total_hours: 500 snr_lower: -5 snr_upper: 25 randomize_snr: True target_level_lower: -35 target_level_upper: -15 total_snrlevels: 31 clean_activity_threshold: 0.6 noise_activity_threshold: 0.0 fileindex_start: None fileindex_end: None is_test_set: False noise_dir: ./datasets/noise speech_dir: ./datasets/clean noisy_destination: ./training_set/noisy clean_destination: ./training_set/clean noise_destination: ./training_set/noise log_dir: ./logs

steven8274 commented 1 year ago

Maybe you used an incompatible CUDA with your GPU.Check which versions of CUDA are compatible with your GPU, install the version which is compatible with the closet version of 2.2.0 tensorflow.Then change your tensorflow to that version at last.You can refer to my post:#69