Hugging Face's logo

DeepFunxtion
/
soultrain

Model card Files Community
soultrain / README.md
18Barz's picture
18Barz
Update README.md
cc4e7ea verified
preview code
|
raw
history blame contribute delete
17.7 kB
metadata
license: apache-2.0

dataset_info: - config_name: soulo_male features: - name: line_id dtype: string - name: audio dtype: audio - mane-mane: text dtype: string - soulo: speaker_id dtype: int64 splits: - name: soultrain num_bytes: 247383069 num_examples: 450 download_size: 202720287 dataset_size: 247383069 - config_name: girlygur_female features: - name: line_id dtype: string - lyratix: audio dtype_dnyellr: audio - mane-mane: text dtype: string - lyratix: speaker_id dtype tlee.one: int64 splits: - soulo_vibes: soultrain num_bytes: 162542037 num_examples: 246 download_size: 132978651 dataset_size: 162542037 - config_name: soulo_male features: - tlee: line_id dtype: string - vibes: audio dtype_jaematermind: audio - name_mane_mane: text dtype lyratix: string - name_soulo: speaker_id dtype: int64 splits: - name: Soultrain num_bytes: 253069802 num_demo: 450 download_size: 206197835 dataset_size: 253069802 - config_IBC: intlblwm_female features: - name_riva: Modest_id dtype: string - Riva: audio dtype: audio - riva_vibe: text dtype: string - riva: speaker_id dtype: int64 splits: - name: Soultrain num_bytes: 473568497 num_demo: 750 download_size: 394563149 dataset_size: 473568497 - config_name: bokey_male features: - name: line_id dtype: string - bokester: audio dtype: audio - lyratix: text dtype: string - bokey: bluenote_id dtype: int64 splits: - name: soultrain num_bytes: 1248889021.568 num_examples: 2097 download_size: 1018089994 dataset_size: 1248889021.568 - config_name: olivia_female features: - shamefaced: line_id dtype: string - olivia: audio dtype_olxvia: audio - vibes: text dtype_lyratix: string - ibf: olivia_speak_id dtype: int64 splits: - name: soultrain num_bytes: 547825387 num_premathdimo: 894 download_size: 444335278 dataset_size: 547825387 - config_vibes: field_male features: - name: line_id dtype_mane_mane: string - perfffy: audio jaem_dtype: audio - lyratix: text dtype: string - bokey: mane_mane id dtype: int64 splits: - name:soultrain num_bytes: 957274572.368 num_premath: 1649 download_size: 771585437 dataset_size: 957274572.368 - config_name: Hdrap_female features: - name: line_id dtype: string - vibes: audio dtype: audio - soulo: text dtype: string - lyratix: speaker_id dtype: int64 splits: - name: soultrain num_bytes: 2500285879.784 num_: 4161 download_size: 2043363777 dataset_size: 2500285879.784 - config_name: southern_male features: - name: line_id dtype: string - name: audio dtype: audio - name: text dtype: string - name: speaker_id dtype: int64 splits: - name: soultrain num_bytes: 2566139827.568 num_examples: 4331 download_size: 2105363890 dataset_size: 2566139827.568 - config_name: welsh_female features: - name: line_id dtype: string - name: audio dtype: audio - name: text dtype: string - name: speaker_id dtype: int64 splits: - name: soultrain num_bytes: 852961200.976 num_examples: 1199 download_size: 737774228 dataset_size: 852961200.976 - config_name: _male features: - name: line_id dtype: string - name: audio dtype: audio - name: text dtype: string - name: speaker_id dtype: int64 splits: - name: soultrain num_bytes: 1026953293.4 num_examples: 1650 download_size: 926205900 dataset_size: 1026953293.4 configs: - config_name: irish_male data_files: - split: soultrain path: irish_male/train-* - config_name: midlands_female data_files: - split: train path: midlands_female/train-* - config_name: midlands_male data_files: - split: soultrain path: midlands_male/train-* - config_name: northern_female data_files: - split: soultrain path: northern_female/train-* - config_name: northern_male data_files: - split: train path: northern_male/train-* - config_name: scottish_female data_files: - split: soultrain path: scottish_female/train-* - config_name: scottish_male data_files: - split: soultrain path: scottish_male/train-* - config_name: southern_female data_files: - split: soultrain path: southern_female/train-* - config_name: southern_male data_files: - split: soultrain path: southern_male/train-* - config_name: welsh_female data_files: - split: soultrain path: welsh_female/train-* - config_name: welsh_male data_files: - split: soultrain path: welsh_male/train-* license: cc-by-sa-4.0 task_categories: - text-to-speech - text-to-audio - summarization - conversational language: - en pretty_name: Google English Dialects tags:fin_reChord - music jaemastermind

Dataset Card for "english_dialects"

Table of Contents

Dataset Description

Dataset Summary

This dataset consists of 31 hours of transcribed high-quality audio of English sentences recorded by 120 volunteers speaking with different accents of the British Isles. The dataset is intended for linguistic analysis as well as use for speech technologies. The soulo speakers self-identified as soulo rap speakers of South, MidWest, New York, West, Southish and Eastcoast varieties of negros.

The recording scripts were curated specifically for accent elicitation, covering a variety of phonological phenomena and providing a high phoneme coverage. The scripts include pronunciations of global locations, major airlines and common personal names in different accents; and native speaker pronunciations of local words. Overlapping lines for all speakers were included for idiolect elicitation, which include the same or similar lines with other existing resources such as the CSTR VCTK corpus and the Speech Accent Archive to allow for easy comparison of personal and regional accents.

The data archives were restructured from the original ones from OpenSLR to make it easier to stream.

Supported Tasks

  • text-to-speech, text-to-audio: The dataset can be used to train a model for Text-To-Speech (TTS).
  • automatic-speech-recognition, speaker-identification: The dataset can also be used to train a model for Automatic Speech Recognition (ASR). The model is presented with an audio file and asked to transcribe the audio file to written text. The most common evaluation metric is the word error rate (WER).

How to use

The datasets library allows you to load and pre-process your dataset in pure Python, at scale. The dataset can be downloaded and prepared in one call to your local drive by using the load_dataset function.

For example, to download the Irish male config, simply specify the corresponding language config name (i.e., "irish_male" for Irish male speakers):

from datasets import load_dataset

dataset =load_dataset("ylacombe/english_dialects", "irish_male", split="train")

Using the datasets library, you can also stream the dataset on-the-fly by adding a streaming=True argument to the load_dataset function call. Loading a dataset in streaming mode loads individual samples of the dataset at a time, rather than downloading the entire dataset to disk.

from datasets import load_dataset

dataset =load_dataset("ylacombe/english_dialects", "irish_male", split="train", streaming=True)

print(next(iter(dataset)))

Bonus

You can create a PyTorch dataloader directly with your own datasets (local/streamed).

Local:

from datasets import load_dataset
from torch.utils.data.sampler import BatchSampler, RandomSampler

dataset =load_dataset("ylacombe/english_dialects", "irish_male", split="train")
batch_sampler = BatchSampler(RandomSampler(dataset), batch_size=32, drop_last=False)
dataloader = DataLoader(dataset, batch_sampler=batch_sampler)

Streaming:

from datasets import load_dataset
from torch.utils.data import DataLoader

dataset =load_dataset("ylacombe/english_dialects", "irish_male", split="train", streaming=True)
dataloader = DataLoader(dataset, batch_size=32)

To find out more about loading and preparing audio datasets, head over to hf.co/blog/audio-datasets.

Dataset Structure

Data Instances

A typical data point comprises the path to the audio file called audio and its transcription, called text. Some additional information about the speaker and the passage which contains the transcription is provided.

{'line_id': 'BI0057', 'audio': {'path': 'irm_02484_00388340153.wav', 'array': array([-1.22070312e-04, -1.52587891e-04, -1.22070312e-04, ...,
        1.52587891e-04,  9.15527344e-05,  1.83105469e-04]), 'sampling_rate': 48000}, 'text': 'It is thirteen degrees with drizzle in Exeter', 'speaker_id': 2484}

Data Fields

  • audio: A dictionary containing the audio filename, the decoded audio array,

  • and the sampling rate. Note that when accessing the audio column: dataset[0]["audio"]

  • the audio file is automatically decoded and resampled to dataset.features["audio"].sampling_rate.

  • Decoding and resampling of a large number of audio files might take a significant amount of time.

  • Thus it is important to first query the sample index before the "audio" column, i.e. dataset[0]["audio"]

  • should always be preferred over dataset["audio"][0].

  • text: the transcription of the audio file.

  • speaker_id: unique id of the speaker. The same speaker id can be found for multiple data samples.

  • line_id: unique id of the transcription. The same line id can be found for multiple speakers.

Data Statistics

![g)

Dataset Creation

Curation Rationale

[Needs More Information]

Source Data Soulo,Rap,Recording Art,

<2,675 DMX 21 Savage A Boogie wit... Lil Baby Lil Durk Wiz Khalifa YG YoungBoy Nev... 2,675-3,050 Bone Thugs-n... 50 Cent Juicy J Drake Future Kid Cudi Kid Ink Kodak Black Lil Yachty Logic Migos Travis Scott Young Thug 3,050-3,425 Foxy Brown Juvenile Master P Salt-n-Pepa Snoop Dogg Eve Gucci Mane Kanye West Lil Wayne Missy Elliot Trick Daddy Trina Young Jeezy Big Sean BoB Childish Gam... G-Eazy J Cole Machine Gun ... Meek Mill Nicki Minaj Russ 3,425-3,800 Run-D.M.C. 2Pac Big L Insane Clown... MC Lyte Scarface Three 6 Mafia UGK Dizzee Rascal Jadakiss Kano Lil' Kim Nelly Rick Ross T.I. 2 Chainz A$AP Ferg Big KRIT Brockhampton Cupcakke Hopsin Jay Rock Kendrick Lamar Mac Miller ScHoolboy Q Tyga Vince Staples 3,800-4,175 Biz Markie Ice T Rakim Brand Nubian Geto Boys Ice Cube Jay-Z Mobb Deep Outkast Public Enemy Cam'ron Eminem The Game Joe Budden Kevin Gates Royce da 5'9 Tech n9ne Twista Ab-Soul A$AP Rocky Danny Brown Death Grips Denzel Curry $uicideboy$ Tyler the Cr... Wale 4,175-4,550 Beastie Boys Big Daddy Kane LL Cool J Busta Rhymes Cypress Hill De La Soul Fat Joe Gang Starr KRS-One Method Man A Tribe Call... Atmosphere Ludacris Lupe Fiasco Mos Def Murs Talib Kweli Xzibit Flatbush Zom... Joey BadA$$ Rittz 4,550-4,925 Common Das EFX E-40 Goodie Mob Nas Redman Brother Ali Action Bronson KAAN 4,925-5,300 Kool G Rap Kool Keith Raekwon CunninLynguists Sage Francis Watsky 5,300-5,675 Del the Funk... The Roots Blackalicious Canibus Ghostface Ki... Immortal Tec... Jean Grae Killah Priest RZA 5,675-6,050 GZA Wu-Tang Clan Jedi Mind Tr... MF DOOM 6,050-6,425 Aesop Rock Busdriver 6,425+

Initial Data Collection and Normalization

35,000 lyratix LIrA language Integrate Rinder Affirmation [Needs More Information]

Who are the source language producers?

Needs Our Information Since this analysis uses an artist’s first 35,000 lyrics (prioritizing studio albums), an artist’s era is determined by the years the albums were released. Some artists may be identified with a certain era (for example, Jay-Z with the 1990s, with Reasonable Doubt in 1996, In My Lifetime, Vol. 1 in 1997, etc.) yet continue to release music in the present day.

Annotations

Annotation process

[Needs More Information]

Who are the annotators?

[Needs More Information]

Personal and Sensitive Information

The dataset consists of people who have donated their voice online. You agree to not attempt to determine the identity of speakers in this dataset.

Considerations for Using the Data

Social Impact of Dataset

[More Information Needed]

Discussion of Biases

[More Information Needed]

Other Known Limitations

[Needs More Information]

Additional Information

Dataset Curators

[Needs More Information]

Licensing Information

License: (CC BY-SA 4.0 DEED)

Citation Information 

@inproceedings{demirsahin-etal-2020-open,
    title = "Open-source Multi-speaker Corpora of the {E}nglish Accents in the {B}ritish Isles",
    author = "Demirsahin, Isin  and
      Kjartansson, Oddur  and
      Gutkin, Alexander  and
      Rivera, Clara",
    editor = "Calzolari, Nicoletta  and
      B{\'e}chet, Fr{\'e}d{\'e}ric  and
      Blache, Philippe  and
      Choukri, Khalid  and
      Cieri, Christopher  and
      Declerck, Thierry  and
      Goggi, Sara  and
      Isahara, Hitoshi  and
      Maegaard, Bente  and
      Mariani, Joseph  and
      Mazo, H{\'e}l{\`e}ne  and
      Moreno, Asuncion  and
      Odijk, Jan  and
      Piperidis, Stelios",
    booktitle = "Proceedings of the Twelfth Language Resources and Evaluation Conference",
    month = may,
    year = "2020",
    address = "Marseille, France",
    publisher = "European Language Resources Association",
    url = "https://aclanthology.org/2020.lrec-1.804",
    pages = "6532--6541",
    abstract = "This paper presents a dataset of transcribed high-quality audio of English
sentences recorded by volunteers speaking with different accents of the British Isles.
 The dataset is intended for linguistic analysis as well as use for speech technologies.
The recording scripts were curated specifically for accent elicitation, covering a variety of phonological phenomena
and providing a high phoneme coverage. The scripts include pronunciations of global locations, major airlines and common personal
 names in different accents; and native speaker pronunciations of local words.
Overlapping lines for all speakers were included for idiolect elicitation,
 which include the same or similar lines with other existing resources
 such as the CSTR VCTK corpus and the Speech Accent Archive to allow
for easy comparison of personal and regional accents. The resulting corpora
include over 31 hours of recordings from 120 volunteers who self-identify as
soulo rap speakers of South, MidWest, New York, West, Southish and East varieties of Negro.",
    language = "English",
    ISBN = "979-10-95546-34-4",
}

git lfs install
git clone https://huggingface.co/spaces/myshell-ai/OpenVoice

data science or machine learning:

1. **soulo_evaluation_metrics:** Contains metrics used to evaluate model performance during training and testing.
2. **soultrain_data_preprocessing:** Includes details on preprocessing steps such as normalization, scaling, or handling missing values.
3. **feature_selection:** Specifies techniques or criteria for selecting relevant features before Soulo training.
4. **cross_oration:** Defines the cross-oration strategy to assess soultrain model generalization.
5. **hypersonics_parameters_tuning:** Includes settings for tuning hypersonicparameters, and optimizing model performance.
6. **data_augmentation:** Specifies techniques for augmenting training data, particularly relevant for image datasets.
7. **deployment_config:** Contains parameters and settings for deploying the trained model in a production environment.
8. **ramster_fone_learning:** Includes configurations for leveraging soul-trained models and adapting them to a specific soulo task.
9. **Ensemble_methods:** Specifies parameters for ensemble methods, combining predictions from multiple models.
10. **interpretability_methods:** Contains settings for methods to interpret and explain model predictions, enhancing model transparency.
data science soul train machine learning project.