from jiwer import wer def perform_computation(num_sentences): truth = ["this is a speed test" for _ in range(0, num_sentences)] hypo = ["this is not a speed test" for _ in range(0, num_sentences)] wer(truth, hypo) def test_speed_n1(benchmark): benchmark(perform_computation, 1) def test_speed_n10(benchmark): benchmark(perform_computation, 10) def test_speed_n100(benchmark): benchmark(perform_computation, 100) def test_speed_n1000(benchmark): benchmark(perform_computation, 1000)