|
--- |
|
datasets: |
|
- AliMeeting |
|
language: |
|
- zh |
|
license: apache-2.0 |
|
metrics: |
|
- |
|
name: "IHM test CER" |
|
type: cer |
|
value: 11.53 |
|
- |
|
name: "SDM test CER" |
|
type: cer |
|
value: 25.85 |
|
- |
|
name: "GSS test CER" |
|
type: cer |
|
value: 14.22 |
|
tags: |
|
- k2 |
|
- icefall |
|
--- |
|
|
|
# AliMeeting |
|
|
|
This is an ASR recipe for the AliMeeting corpus. AliMeeting provides recordings from the speaker's |
|
headset microphones and an 8-channel microphone array. |
|
We pool data in the following 4 ways and train a single model on the pooled data: |
|
|
|
(i) individual headset microphone (IHM) |
|
(ii) IHM with simulated reverb |
|
(iii) Single distant microphone (SDM) |
|
(iv) GSS-enhanced array microphones |
|
|
|
Speed perturbation and MUSAN noise augmentation are additionally performed on the pooled |
|
data. |
|
|
|
## Performance Record |
|
|
|
### pruned_transducer_stateless7 |
|
|
|
The following are decoded using `modified_beam_search`: |
|
|
|
| Evaluation set | eval CER | test CER | |
|
|--------------------------|------------|---------| |
|
| IHM | 9.58 | 11.53 | |
|
| SDM | 23.37 | 25.85 | |
|
| MDM (GSS-enhanced) | 11.82 | 14.22 | |
|
|
|
See the [recipe](https://github.com/k2-fsa/icefall/tree/master/egs/alimeeting/ASR_v2) for details. |
|
|
