import argparse import os import torch from omnilingual_asr.models.inference.pipeline import ASRInferencePipeline import evaluate from normalizer import data_utils import time from tqdm import tqdm wer_metric = evaluate.load("wer") def main(args): # Map model_id to model_card format expected by omnilingual_asr # e.g., "facebook/omniASR-LLM-7B" -> "omniASR_LLM_7B" model_card = args.model_id.split("/")[-1].replace("-", "_") # Initialize the ASR pipeline # Convert device integer to torch.device object if args.device >= 0: device = torch.device(f"cuda:{args.device}") else: device = torch.device("cpu") pipeline = ASRInferencePipeline( model_card=model_card, device=device ) MAX_AUDIO_SEC = 40 # Pipeline max audio length (see omnilingual_asr MAX_ALLOWED_AUDIO_SEC) def benchmark(batch): # Load audio inputs minibatch_size = len(batch["audio"]) batch["audio_length_s"] = [len(audio["array"]) / audio["sampling_rate"] for audio in batch["audio"]] batch["audio_filepath"] = data_utils.extract_audio_filepaths_from_batch(batch, minibatch_size) # Convert to pipeline input format: list of dicts with waveform and sample_rate # Truncate audio to MAX_AUDIO_SEC to avoid pipeline assert_max_length errors audio_data = [] for audio in batch["audio"]: waveform = audio["array"] sample_rate = audio["sampling_rate"] max_samples = int(MAX_AUDIO_SEC * sample_rate) if len(waveform) > max_samples: waveform = waveform[:max_samples] audio_data.append({"waveform": waveform, "sample_rate": sample_rate}) # START TIMING start_time = time.time() lang = [args.language] * minibatch_size transcriptions = pipeline.transcribe( audio_data, lang=lang, batch_size=minibatch_size ) # END TIMING runtime = time.time() - start_time # normalize by minibatch size since we want the per-sample time batch["transcription_time_s"] = minibatch_size * [runtime / minibatch_size] batch["predictions"] = transcriptions # raw; normalization applied at scoring time batch["references"] = batch["original_text"] # raw; normalization applied at scoring time return batch if args.warmup_steps is not None: dataset = data_utils.load_data(args) dataset = data_utils.prepare_data(dataset) num_warmup_samples = args.warmup_steps * args.batch_size if args.streaming: warmup_dataset = dataset.take(num_warmup_samples) else: warmup_dataset = dataset.select(range(min(num_warmup_samples, len(dataset)))) warmup_dataset = iter(warmup_dataset.map(benchmark, batch_size=args.batch_size, batched=True)) for _ in tqdm(warmup_dataset, desc="Warming up..."): continue dataset = data_utils.load_data(args) if args.max_eval_samples is not None and args.max_eval_samples > 0: print(f"Subsampling dataset to first {args.max_eval_samples} samples!") if args.streaming: dataset = dataset.take(args.max_eval_samples) else: dataset = dataset.select(range(min(args.max_eval_samples, len(dataset)))) dataset = data_utils.prepare_data(dataset) dataset = dataset.map( benchmark, batch_size=args.batch_size, batched=True, remove_columns=["audio"], ) all_results = { "audio_length_s": [], "transcription_time_s": [], "predictions": [], "references": [], "audio_filepath": [], } result_iter = iter(dataset) for result in tqdm(result_iter, desc="Samples..."): for key in all_results: all_results[key].append(result[key]) # Write manifest results (WER and RTFX) manifest_path = data_utils.write_manifest( all_results["references"], all_results["predictions"], args.model_id, args.dataset_path, args.dataset, args.split, audio_length=all_results["audio_length_s"], transcription_time=all_results["transcription_time_s"], audio_filepaths=all_results["audio_filepath"], ) print("Results saved at path:", os.path.abspath(manifest_path)) norm_refs = [data_utils.normalizer(r) for r in all_results["references"]] norm_preds = [data_utils.normalizer(p) for p in all_results["predictions"]] wer = wer_metric.compute( references=norm_refs, predictions=norm_preds ) wer = round(100 * wer, 2) rtfx = round(sum(all_results["audio_length_s"]) / sum(all_results["transcription_time_s"]), 2) print("WER:", wer, "%", "RTFx:", rtfx) if __name__ == "__main__": parser = argparse.ArgumentParser() parser.add_argument( "--model_id", type=str, required=True, help="Model identifier on Hugging Face (e.g., 'facebook/omniASR-LLM-7B')", ) parser.add_argument( "--dataset_path", type=str, default="hf-audio/open-asr-leaderboard", help="Dataset path. By default, it is `hf-audio/open-asr-leaderboard`", ) parser.add_argument( "--dataset", type=str, required=True, help="Dataset name. *E.g.* `'librispeech_asr` for the LibriSpeech ASR dataset, or `'common_voice'` for Common Voice. The full list of dataset names " "can be found at `https://huggingface.co/datasets/hf-audio/open-asr-leaderboard`", ) parser.add_argument( "--split", type=str, default="test", help="Split of the dataset. *E.g.* `'validation`' for the dev split, or `'test'` for the test split.", ) parser.add_argument( "--device", type=int, default=-1, help="The device to run the pipeline on. -1 for CPU (default), 0 for the first GPU and so on.", ) parser.add_argument( "--batch_size", type=int, default=16, help="Number of samples to go through each streamed batch.", ) parser.add_argument( "--max_eval_samples", type=int, default=None, help="Number of samples to be evaluated. Put a lower number e.g. 64 for testing this script.", ) parser.add_argument( "--streaming", action="store_true", help="Stream the dataset lazily over the network instead of downloading it in full before the evaluation. Off by default for reproducible benchmark timings.", ) parser.add_argument( "--warmup_steps", type=int, default=2, help="Number of warm-up steps to run before launching the timed runs.", ) parser.add_argument( "--language", type=str, default="eng_Latn", help="Language code for transcription (e.g., 'eng_Latn' for English). Set to None for language-agnostic transcription.", ) args = parser.parse_args() main(args)