YAML Metadata
Warning:
empty or missing yaml metadata in repo card
(https://huggingface.co/docs/hub/model-cards#model-card-metadata)
Wenet ASR cpu runtime benchmark
AliCould Instance: ecs.c7.16xlarge 64 vCPU 128 GiB 2.7 GHz/3.5 GHz 32 Gbps, Intel Xeon(Ice Lake) Platinum 8369B
Offline ASR
Num Concurrent Tasks: 256 (Best among 16, 32, 64, 128, 256, 512)
| Model | Aishell1 Test WER (%) | 10 Hours Decoding Time (secs) |
|---|---|---|
| Aishell1_small_offline_fp32_ctc_prefix | 4.62 | 74.56 |
| Aishell1_small_offline_int8_ctc_prefix | 4.79 | 59.08 |
| Aishell1_small_offline_fp32_ctc_wfst | 4.38 | 90.91 |
| Aishell1_small_offline_int8_ctc_wfst | 4.50 | 69.57 |
| WenetSpeech_large_offline_fp32_ctc_prefix | 4.61 | 181.77 |
| WenetSpeech_large_offline_int8_ctc_prefix | 4.71 | 125.57 |
Streaming ASR
Num Concurrent Tasks: 256 (Best among 16, 32, 64, 128, 256, 512)
| Model | Aishell1 Test WER (%) | 10 Hours Decoding Time (secs) |
|---|---|---|
| Aishell1_small_u2pp_fp32_ctc_prefix | 102.21 | |
| Aishell1_small_u2pp_fp32_ctc_wfst | 106.67 | |
| WenetSpeech_large_u2pp_fp32_ctc_prefix | 276.32 |
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support