Dataset Viewer
Auto-converted to Parquet Duplicate
audio
audioduration (s)
2.78
9.34
transcript
stringlengths
27
80
language
stringclasses
3 values
script
stringclasses
3 values
voice
stringclasses
6 values
duration_s
float64
2.78
9.34
sample_rate
int64
16k
16k
channels
int64
1
1
rms
float64
0.09
0.16
source
stringclasses
1 value
synthetic
bool
1 class
Hello! My name is Sadik. Welcome to the SadikAI Multiverse audio dataset.
en
Latin
en-US-AriaNeural
7.824
16,000
1
0.12269
Microsoft neural TTS via edge-tts
true
The sun rises in the east and sets in the west.
en
Latin
en-US-GuyNeural
3.504
16,000
1
0.14243
Microsoft neural TTS via edge-tts
true
Water boils at one hundred degrees Celsius at sea level.
en
Latin
en-US-AriaNeural
4.224
16,000
1
0.12921
Microsoft neural TTS via edge-tts
true
Please open your books to page twelve.
en
Latin
en-US-GuyNeural
3.024
16,000
1
0.14679
Microsoft neural TTS via edge-tts
true
A triangle has three sides and three angles.
en
Latin
en-US-AriaNeural
3.624
16,000
1
0.1217
Microsoft neural TTS via edge-tts
true
Reading every day helps you learn new words.
en
Latin
en-US-GuyNeural
3.264
16,000
1
0.15745
Microsoft neural TTS via edge-tts
true
The capital of France is Paris.
en
Latin
en-US-AriaNeural
2.856
16,000
1
0.08548
Microsoft neural TTS via edge-tts
true
Be kind, be curious, and never stop asking questions.
en
Latin
en-US-GuyNeural
4.344
16,000
1
0.13049
Microsoft neural TTS via edge-tts
true
Two plus three equals five.
en
Latin
en-US-AriaNeural
2.784
16,000
1
0.08547
Microsoft neural TTS via edge-tts
true
Thank you for listening. Have a wonderful day.
en
Latin
en-US-GuyNeural
4.248
16,000
1
0.13569
Microsoft neural TTS via edge-tts
true
नमस्ते! मेरा नाम सादिक है। सादिक एआई मल्टीवर्स ऑडियो डेटासेट में आपका स्वागत है।
hi
Devanagari
hi-IN-SwaraNeural
9.336
16,000
1
0.13054
Microsoft neural TTS via edge-tts
true
सूरज पूर्व में उगता है और पश्चिम में डूबता है।
hi
Devanagari
hi-IN-MadhurNeural
4.944
16,000
1
0.14073
Microsoft neural TTS via edge-tts
true
समुद्र तल पर पानी सौ डिग्री सेल्सियस पर उबलता है।
hi
Devanagari
hi-IN-SwaraNeural
4.848
16,000
1
0.13792
Microsoft neural TTS via edge-tts
true
कृपया अपनी किताबें पृष्ठ बारह पर खोलिए।
hi
Devanagari
hi-IN-MadhurNeural
4.08
16,000
1
0.14811
Microsoft neural TTS via edge-tts
true
त्रिभुज की तीन भुजाएँ और तीन कोण होते हैं।
hi
Devanagari
hi-IN-SwaraNeural
4.128
16,000
1
0.14119
Microsoft neural TTS via edge-tts
true
हर दिन पढ़ने से नए शब्द सीखने में मदद मिलती है।
hi
Devanagari
hi-IN-MadhurNeural
5.016
16,000
1
0.13834
Microsoft neural TTS via edge-tts
true
फ्रांस की राजधानी पेरिस है।
hi
Devanagari
hi-IN-SwaraNeural
3.168
16,000
1
0.14432
Microsoft neural TTS via edge-tts
true
दयालु रहें, जिज्ञासु रहें और सवाल पूछना कभी न छोड़ें।
hi
Devanagari
hi-IN-MadhurNeural
5.256
16,000
1
0.15093
Microsoft neural TTS via edge-tts
true
दो और तीन मिलकर पाँच होते हैं।
hi
Devanagari
hi-IN-SwaraNeural
3.456
16,000
1
0.14725
Microsoft neural TTS via edge-tts
true
सुनने के लिए धन्यवाद। आपका दिन शुभ हो।
hi
Devanagari
hi-IN-MadhurNeural
4.92
16,000
1
0.1563
Microsoft neural TTS via edge-tts
true
سلام! میرا نام صادق ہے۔ صادق اے آئی ملٹی ورس آڈیو ڈیٹاسیٹ میں خوش آمدید۔
ur
Arabic
ur-PK-AsadNeural
8.04
16,000
1
0.12905
Microsoft neural TTS via edge-tts
true
سورج مشرق میں طلوع ہوتا ہے اور مغرب میں غروب ہوتا ہے۔
ur
Arabic
ur-PK-UzmaNeural
4.272
16,000
1
0.14188
Microsoft neural TTS via edge-tts
true
سمندر کی سطح پر پانی سو ڈگری سیلسیس پر ابلتا ہے۔
ur
Arabic
ur-PK-AsadNeural
4.392
16,000
1
0.13101
Microsoft neural TTS via edge-tts
true
براہ کرم اپنی کتابیں صفحہ بارہ پر کھولیں۔
ur
Arabic
ur-PK-UzmaNeural
3.48
16,000
1
0.14636
Microsoft neural TTS via edge-tts
true
مثلث کی تین ضلعیں اور تین زاویے ہوتے ہیں۔
ur
Arabic
ur-PK-AsadNeural
3.72
16,000
1
0.14228
Microsoft neural TTS via edge-tts
true
ہر روز پڑھنے سے نئے الفاظ سیکھنے میں مدد ملتی ہے۔
ur
Arabic
ur-PK-UzmaNeural
3.888
16,000
1
0.14421
Microsoft neural TTS via edge-tts
true
فرانس کا دارالحکومت پیرس ہے۔
ur
Arabic
ur-PK-AsadNeural
3
16,000
1
0.14799
Microsoft neural TTS via edge-tts
true
مہربان رہیں، متجسس رہیں اور سوال پوچھنا کبھی نہ چھوڑیں۔
ur
Arabic
ur-PK-UzmaNeural
4.344
16,000
1
0.13738
Microsoft neural TTS via edge-tts
true
دو اور تین مل کر پانچ ہوتے ہیں۔
ur
Arabic
ur-PK-AsadNeural
2.88
16,000
1
0.13538
Microsoft neural TTS via edge-tts
true
سننے کے لیے شکریہ۔ آپ کا دن اچھا گزرے۔
ur
Arabic
ur-PK-UzmaNeural
4.656
16,000
1
0.14052
Microsoft neural TTS via edge-tts
true

SadikAI Audio Multiverse

30 short synthetic speech clips with exact transcripts: 10 in English, 10 in Hindi, and 10 in Urdu. Clips are 2.8–9.3 seconds long, mono, 16 kHz PCM WAV. Six different neural voices are used (two per language).

These are text-to-speech clips, not human recordings. The audio was generated with Microsoft's online neural voices through the edge-tts client. It is clean, read-style speech with no background noise. Do not use it for voice cloning, impersonation, or claims about real speakers.

language clips duration voices
English (en) 10 39.7 s Aria, Guy
Hindi (hi) 10 49.2 s Swara, Madhur
Urdu (ur) 10 42.6 s Asad, Uzma
total 30 131.5 s 6

Files

data/
  metadata.jsonl
  en/*.wav
  hi/*.wav
  ur/*.wav

Each row in metadata.jsonl has:

field meaning
file_name audio path relative to data/
transcript exact text spoken in the clip
language en, hi, or ur
script Latin, Devanagari, or Arabic
voice exact Microsoft voice name
duration_s duration in seconds
sample_rate 16000
channels 1
rms measured loudness proxy after normalisation
source Microsoft neural TTS via edge-tts
synthetic always true

Load it

from datasets import load_dataset

ds = load_dataset("audiofolder", data_dir="data")
print(ds)
print(ds["train"][0])

Or read the metadata as JSON:

from datasets import load_dataset
ds = load_dataset("json", data_files="data/metadata.jsonl")

What is in the clips

The transcripts are simple, school-friendly sentences about greetings, the sun, water boiling, books, triangles, reading, Paris, kindness, arithmetic, and saying thanks. They are deliberately short and easy to align.

How it was made

  • Text written for this dataset.
  • Speech generated with edge-tts (Microsoft neural voices).
  • Converted to 16 kHz mono PCM WAV with ffmpeg and loudness-normalised to roughly −16 LUFS.
  • Every clip was checked for non-silent audio and a duration above 1 second.

Licensing and attribution

  • Transcripts, metadata, and packaging: CC BY 4.0 (see LICENSE).
  • Audio: synthetic output from Microsoft neural TTS. The edge-tts client is an unofficial interface to Microsoft's online speech service. Review Microsoft's terms before commercial use or redistribution outside this dataset.
  • No human speakers were recorded. No personal data is included.
  • If you reuse the audio, credit Sadik Nawaz (SadikAI), SadikAI Audio Multiverse.

Limitations

  • Only 30 clips and 6 voices; not a speaker-recognition or ASR benchmark.
  • Read speech only: no noise, overlap, emotion, or spontaneous conversation.
  • Hindi and Urdu are read by neural voices with limited accent variety; some names and loanwords may be pronounced differently than a local speaker would.
  • Loudness normalisation changes the original TTS level.
  • Transcripts are exact, but punctuation and pauses are not annotated.

Credits

Created by Sadik Nawaz (SadikAI) as part of the SadikAI Multiverse dataset family. Voices are provided by Microsoft's neural TTS service.

Downloads last month
64

Collection including SadikNawazAI/sadik-audio-multiverse