jml2026/split_sample

Silencio Network: Multilingual Accent Speech Dataset (Sample)

Overview

This dataset is a crowdsourced multilingual–accented English and non-English speech dataset designed for model training, benchmarking, and acoustic analysis. It emphasizes accent variation, short-form scripted prompts, and spontaneous free speech. All recordings were produced by contributors using their own devices, with Whisper-generated transcripts provided for every sample.

The dataset is structured for direct use in ASR, TTS, accent-classification, diarization-adjacent analysis, speech segmentation, and embedding evaluation.

Languages and Accents

This dataset covers five language–region pairs (to find out more about other combinations please reach out to us):

• English (China): English spoken with Mandarin-influenced accent
• English (Nigeria): Nigerian-accented English
• English (United States): American English
• German (Germany): Native German speakers
• Spanish (Mexico): Native Mexican Spanish speakers

All recordings are stored as 48 kHz WAV files.

Speech Types

Each sample belongs to one of three categories:

• free_speech: unscripted speech on a provided topic
• keywords: short isolated prompts containing specific phrases or terms
• monologues: longer scripted passages

These values appear in the field type_of_script.

Recording Conditions

All data is crowdsourced. Contributors record themselves using their available hardware and environment; conditions therefore vary naturally across microphones, devices, and noise profiles. No studio-grade normalisation or homogenisation is applied.

Transcription

Transcriptions are machine-generated using OpenAI Whisper, preserving its segmentation structure where applicable.

Dataset Statistics

Durations are given in hours. Counts reflect samples within each (language, region, type_of_script) partition.

English (China)

type_of_script	duration_hrs	recordings	speakers
free_speech	0.99	72	19
keywords	0.48	57	10
monologues	0.98	56	11

English (Nigeria)

type_of_script	duration_hrs	recordings	speakers
free_speech	0.98	75	65
keywords	0.99	141	101
monologues	0.99	49	32

English (United States)

type_of_script	duration_hrs	recordings	speakers
free_speech	0.99	80	35
keywords	0.99	119	40
monologues	0.99	78	27

German (Germany)

type_of_script	duration_hrs	recordings	speakers
free_speech	0.98	99	34
keywords	0.99	152	37
monologues	0.98	77	27

Spanish (Mexico)

type_of_script	duration_hrs	recordings	speakers
free_speech	0.98	90	6
keywords	0.05	6	2
monologues	0.70	45	9

File Structure

data/
    english_china/
        train-0000.parquet
    english_nigeria/
        train-0000.parquet
    english_united_states/
        train-0000.parquet
    german_germany/
        train-0000.parquet
    spanish_mexico/
        train-0000.parquet

Each parquet contains a mixture of free_speech, keywords, and monologues.

Feature Schema

All configurations share the same feature structure:

• id: integer (unique identifier)
• speaker_id: string (hashed or anonymized speaker ID)
• gender: string (speaker gender)
• ethnicity: string (speaker ethnicity)
• occupation: float (occupation or profession, stored as float per original schema)
• country_code: string (ISO 3166-1 alpha-2 code)
• birth_place: string (country or region of birth)
• mother_tongue: string (native language)
• dialect: string (regional dialect)
• year_of_birth: int (birth year, YYYY)
• years_at_birth_place: int (years lived at birth place)
• languages_data: string (serialized language–proficiency data)
• os: string (recording operating system)
• device: string (recording device type)
• browser: string (browser used if web-based)
• duration: float (seconds) (audio length)
• emotions: string (brace-formatted emotion labels)
• language: string (primary language of the recording)
• location: string (recording location category)
• noise_sources: string (brace-formatted background noise labels)
• script_id: int (script template identifier)
• type_of_script: string {free_speech, keywords, monologues} (script category)
• script: string (text intended to be spoken)
• transcript: string (Whisper-generated transcription)
• transcription_segments: string (serialized segmentation with timing and word data)
• audio: WAV audio object (associated audio file)

Licensing

Released under CC BY-NC 4.0.
Commercial use is not permitted. Attribution to Silencio Network is required for any publication or derivative dataset.

Intended Use

Suitable for:

• accent-conditioned ASR training
• multilingual speech recognition
• TTS voicebank generation
• speaker embedding and similarity evaluation
• robustness benchmarking
• keyword-spotting models
• segmentation and VAD evaluation

Limitations

• Transcripts are automatically generated. Errors may be present.
• Crowdsourced device diversity introduces variable noise levels.

Citation

@dataset{silencio_network_speech_2025,
    title        = {Silencio Network Multilingual Accent Speech Corpus},
    author       = {Silencio Network},
    year         = {2025},
    license      = {CC BY-NC 4.0}
}