Fix training issues (#31)

* Updates the training cargo command

Cargo.toml

@@ -45,7 +45,7 @@ dasp_interpolate = { version = "0.11.0", features = ["sinc"], optional = true }
 dasp_ring_buffer = { version = "0.11.0", optional = true }
 easyfft = "0.3.3"
 glob = { version = "0.3.0", optional = true }
-hdf5 = { version = "0.8.1", optional = true }
+hdf5 = { git = "https://github.com/aldanor/hdf5-rust.git", optional = true }
 hound = { version = "3.4.0", optional = true }
 libc = { version = "0.2.119", optional = true }
 # This needs to be in sync with the version from hdf5; they don't re-export it.

train/README.md

@@ -29,7 +29,7 @@ ffmpeg -i $file -f s16le -ac 1 -ar 48000 output.wav
 Once you have your speech and noise files, you need to use `nnnoiseless` to
 generate training features:
 ```
-cargo run --features=train --bin=train --count=<COUNT> --signal-glob=</PATH/TO/SPEECH/*.wav> --noise-glob=<PATH/TO/NOISE/*.wav> -o training.h5
+cargo run --features=train --bin=train --release -- --count=<COUNT> --signal-glob=</PATH/TO/SPEECH/*.wav> --noise-glob=<PATH/TO/NOISE/*.wav> -o training.h5
 ```
 where `<COUNT>` is the number of frames of training data that you want to
 generate. (I don't know what the optimal number is, but 10 million seems to be