Commit 7213a2060906636344f308cf3eb3879bb086ed01
1 parent
8afb6416be
Exists in
master
Allow user to manage voxceleb files using features files. Other features incoming.
Showing 1 changed file with 70 additions and 0 deletions Side-by-side Diff
volia/voxceleb.py
| 1 | +import argparse | |
| 2 | +from utils import SubCommandRunner | |
| 3 | + | |
| 4 | +def utt2spk(features: str, outfile: str): | |
| 5 | + """Generate a utt2spk file from a feature file of voxceleb. | |
| 6 | + (it also works with list files instead of features) | |
| 7 | + | |
| 8 | + Args: | |
| 9 | + features (str): features file (or list) | |
| 10 | + outfile (str): output file to store the utt2spk | |
| 11 | + """ | |
| 12 | + with open(features, "r") as f, open(outfile, "w") as out: | |
| 13 | + for line in f: | |
| 14 | + splited = line.replace("\n", "").split(" ") | |
| 15 | + id_ = splited[0] | |
| 16 | + id_splited = id_.split("-") | |
| 17 | + spk = id_splited[0] | |
| 18 | + out.write(id_ + " " + spk + "\n") | |
| 19 | + | |
| 20 | + | |
| 21 | +def spk2utt(features: str, outfile: str): | |
| 22 | + """Generate a spk2utt file from a feature file of voxceleb. | |
| 23 | + (it also works with list files instead of features) | |
| 24 | + | |
| 25 | + Args: | |
| 26 | + features (str): features file (or list) | |
| 27 | + outfile (str): output file to store the spk2utt | |
| 28 | + """ | |
| 29 | + with open(features, "r") as f, open(outfile, "w") as out: | |
| 30 | + spk2utt_dict = {} | |
| 31 | + for line in f: | |
| 32 | + splited = line.replace("\n", "").split(" ") | |
| 33 | + id_ = splited[0] | |
| 34 | + id_splited = id_.split("-") | |
| 35 | + spk = id_splited[0] | |
| 36 | + if spk not in spk2utt_dict: | |
| 37 | + spk2utt_dict[spk] = [] | |
| 38 | + spk2utt_dict[spk].append(id_) | |
| 39 | + | |
| 40 | + for spk, ids in spk2utt_dict.items(): | |
| 41 | + out.write(spk + " " + " ".join(ids) + "\n") | |
| 42 | + | |
| 43 | + | |
| 44 | +if __name__ == "__main__": | |
| 45 | + # Main parser | |
| 46 | + parser = argparse.ArgumentParser(description="Voxceleb data management") | |
| 47 | + subparsers = parser.add_subparsers(title="action") | |
| 48 | + | |
| 49 | + # utt2spk | |
| 50 | + parser_utt2spk = subparsers.add_parser("utt2spk", help="Generate utt2spk file from feature file (works with list).") | |
| 51 | + parser_utt2spk.add_argument("--features", required=True, help="Features file (works with list)") | |
| 52 | + parser_utt2spk.add_argument("--outfile", default="utt2spk", help="output file") | |
| 53 | + parser_utt2spk.set_defaults(which="utt2spk") | |
| 54 | + | |
| 55 | + # spk2utt | |
| 56 | + parser_spk2utt = subparsers.add_parser("spk2utt", help="Generate spk2utt file from feature file (works with list).") | |
| 57 | + parser_spk2utt.add_argument("--features", required=True, help="Features file (works with list)") | |
| 58 | + parser_spk2utt.add_argument("--outfile", default="spk2utt", help="output file") | |
| 59 | + parser_spk2utt.set_defaults(which="spk2utt") | |
| 60 | + | |
| 61 | + # Parse | |
| 62 | + args = parser.parse_args() | |
| 63 | + | |
| 64 | + # Run commands | |
| 65 | + runner = SubCommandRunner({ | |
| 66 | + "utt2spk" : utt2spk, | |
| 67 | + "spk2utt": spk2utt, | |
| 68 | + }) | |
| 69 | + | |
| 70 | + runner.run(args.which, args.__dict__, remove="which") |