-
Notifications
You must be signed in to change notification settings - Fork 52
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge pull request #193 from ftshijt/source
Update challenge information
- Loading branch information
Showing
1 changed file
with
28 additions
and
7 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -37,13 +37,22 @@ Participation is open to all. Each team can participate in any task. This challe | |
| - [Singing voice synthesis (SVS)](https://github.com/A-Quarter-Mile/espnet/tree/tmp_muskit/egs2/opencpop/svs2) | ||
| - [Discrete vocoder training](https://github.com/kan-bayashi/ParallelWaveGAN) | ||
|
|
||
| ### Dataset | ||
|
|
||
| ### Track-specific dataset | ||
|
|
||
| - ASR: [Librispeech](https://www.openslr.org/12) and [ML-SUPERB](https://drive.google.com/file/d/1zslKQwadZaYWXAmfBCvlos9BVQ9k6PHT/view?usp=sharing) | ||
| - TTS: [LJSpeech](https://keithito.com/LJ-Speech-Dataset/) and [Expresso](https://speechbot.github.io/expresso/) | ||
| - SVS: [Opencpop](https://wenet.org.cn/opencpop/) | ||
|
|
||
|
|
||
| ### Data for discrete representation learning and extraction | ||
| - General Policy: There are no restrictions on using datasets for learning and extracting discrete representations. This applies broadly to all datasets. | ||
|
|
||
| - Specific Restrictions for Supervision Data: The key restriction is on using test sets from certain datasets for supervision in specific tasks. Specifically: | ||
| - Automatic Speech Recognition (ASR): The test sets of the Librispeech and ML-SUPERB datasets cannot be used for learning the discrete representation. However, their training sets are permissible. | ||
| - Text-to-Speech (TTS): The test sets of the LJSpeech and Expresso datasets are off-limits for discrete representation learning, but their training sets can be used. For TTS training, phone alignment information for non-autoregressive training can be also used in training phase. | ||
| - Singing Voice Synthesis (SVS): The test set of the Opencpop dataset is restricted for use in discrete representation learning, though the training set is allowed. | ||
|
|
||
| <!-- ### Rules | ||
| * For each task, the training data must follow the baseline systems. However, there is no constraint on the data used in the foundation models. | ||
| * For submission, more details will be provided later for each task. | ||
|
|
@@ -55,7 +64,10 @@ Participation is open to all. Each team can participate in any task. This challe | |
|
|
||
| * Data: LibriSpeech_100 + ML-SUPERBB 1h set | ||
| * Framework: We recommend to use ESPnet for fair comparison. Feel free to let us know your preferrence. | ||
| * Evaluation metrics: Word Error Rates (WERs) on 5 test sets. | ||
| * Evaluation metrics: Word Error Rates (WERs) on Librispeech dev/test sets and Character Error Rates (CERs) on ML-SUPERB. | ||
| * Ranking: | ||
| * Word/Character Error Rate: The primary method for ranking all systems is based on their Word/Character Error Rate. This metric measures the performance of a system in terms of the accuracy of the words recognized or generated compared to a reference. | ||
| * Efficiency of discrete tokens (bitrate): In addition to WER, the efficiency of discrete tokens in the systems will also be evaluated and ranked based on bitrate. | ||
| * Submission | ||
| * Submission package details: | ||
| 1. The discrete speech units corresponding to the test sets in kaldi format. | ||
|
|
@@ -65,8 +77,11 @@ Participation is open to all. Each team can participate in any task. This challe | |
| ### TTS Challenge - Acoustic+Vocoder | ||
|
|
||
| * Data: LJSpeech, following the train-dev-test split in [here](https://github.com/ftshijt/Interspeech2024_DiscreteSpeechChallenge). | ||
| * Framework: No framework restriction in TTS-Acoustic+Vocoder challenge, but the organizers have prepared the baseline training scripts (baseline model to be released soon) in [ESPnet](https://github.com/espnet/espnet/tree/tts2/egs2/ljspeech/tts2). | ||
| * Framework: No framework or model restriction in TTS-Acoustic+Vocoder challenge, but the organizers have prepared the baseline training scripts (baseline model to be released soon) in [ESPnet](https://github.com/espnet/espnet/tree/tts2/egs2/ljspeech/tts2). | ||
| * Evaluation metrics: Mean cepstral distortion, F0 root mean square error, Bitrate, [UTMOS](https://github.com/sarulab-speech/UTMOS22/tree/master) | ||
| * Ranking: | ||
| * UTMOS: The The primary method for ranking all systems is based on their UTMOS score. | ||
| * Efficiency of discrete tokens (bitrate): the efficiency of discrete tokens in the systems will also be evaluated and ranked based on bitrate. | ||
| * Submission | ||
| * Submission package details: | ||
| * The synthesized voice of LJSpeech test set using full training set (with at least 16kHz). | ||
|
|
@@ -81,8 +96,11 @@ Participation is open to all. Each team can participate in any task. This challe | |
| ### TTS Challenge - Vocoder | ||
|
|
||
| * Data: Expresso, following the train-dev-test split in [here](https://github.com/ftshijt/Interspeech2024_DiscreteSpeechChallenge) (Note that this is different from the original train-dev-test split in the benchmark paper). | ||
| * Framework: No framework restriction in TTS-Vocoder challenge, but the organizers have prepared the baseline training scripts (baseline model to be released soon) in [ESPnet](https://github.com/espnet/espnet/tree/tts2/egs2/ljspeech/tts2) and [ParallelWaveGAN](https://github.com/kan-bayashi/ParallelWaveGAN). | ||
| * Framework: No framework or model restriction in TTS-Vocoder challenge, but the organizers have prepared the baseline training scripts (baseline model to be released soon) in [ESPnet](https://github.com/espnet/espnet/tree/tts2/egs2/ljspeech/tts2) and [ParallelWaveGAN](https://github.com/kan-bayashi/ParallelWaveGAN). | ||
| * Evaluation metrics: Mean cepstral distortion, F0 root mean square error, Bitrate, [UTMOS](https://github.com/sarulab-speech/UTMOS22/tree/master) | ||
| * Ranking: | ||
| * UTMOS: The The primary method for ranking all systems is based on their UTMOS score. | ||
| * Efficiency of discrete tokens (bitrate): the efficiency of discrete tokens in the systems will also be evaluated and ranked based on bitrate. | ||
| * Submission | ||
| * Submission package details: | ||
| * The synthesized voice of LJSpeech test set using full training set (with at least 16kHz). | ||
|
|
@@ -96,10 +114,13 @@ Participation is open to all. Each team can participate in any task. This challe | |
| ### SVS Challenge | ||
|
|
||
| * Data: Opencpop, following the original segmentation and train/test split. | ||
| * Framework: No framework restriction in SVS challenge, but the organizers have prepared the baseline training scripts (baseline model to be released soon) in [ESPnet-Muskits](https://github.com/A-Quarter-Mile/espnet/tree/tmp_muskit/egs2/opencpop/svs2). | ||
| * Framework: No framework or model restriction in SVS challenge, but the organizers have prepared the baseline training scripts (baseline model to be released soon) in [ESPnet-Muskits](https://github.com/A-Quarter-Mile/espnet/tree/tmp_muskit/egs2/opencpop/svs2). | ||
| * Evaluation metrics | ||
| * Objective metrics: Mean cepstral distortion, F0 root mean square error, Bitrate for efficiency measure | ||
| * Subjective metrics: Mean Opinion Score by organizers | ||
| * Subjective metrics: Mean Opinion Score (MOS) by organizers | ||
| * Ranking: | ||
| * MOS: The The primary method for ranking all systems is based on their MOS score. | ||
| * Efficiency of discrete tokens (bitrate): the efficiency of discrete tokens in the systems will also be evaluated and ranked based on bitrate. | ||
| * Submission | ||
| * Submission package details: | ||
| * The synthesized voice of Opencpop test set (with at least 16kHz) | ||
|
|
@@ -153,4 +174,4 @@ The schedule for the challenge is as follows | |
| * Qin Jin (Renmin University of China, China) | ||
|
|
||
| ## Contact | ||
| - [email protected] | ||
| - [email protected] | ||