Skip to data content

Avlys OTS Data / Audio data

Audio training data for speech AI

Avlys Data helps speech-AI teams scope audio, transcription and review for speech recognition, keyword spotting and model evaluation. The catalogue includes multilingual read speech, Indian-language speech and voice commands. These are not interchangeable with support-call recordings.

Request sample & pricingExplore collections

From Hyderabad and Bengaluru, India. Working with teams globally. Availability confirmed for your brief.

Explore the specifications

Audio data collections

Compare 6 options by task, coverage and format. Request sample availability and a quote for your chosen scope.

Multilingual read speech

Avlys OTS Data

Multilingual read speech

Recorded readings of parallel prompts for multilingual speech recognition and cross-language evaluation.

Reference collection scope102 languages

Read speech · 16 kHz audio

  • Multilingual ASR
  • Speech evaluation
Dataset specifications for Multilingual read speech
Formats & labels
Audio with text and language metadata. Reference transcriptions and published train / validation / test splits.
Fit & scope
Read prompts, not spontaneous conversations or call-centre recordings. Confirm locale coverage and speaker splits for the task.
Request a quote by email: Multilingual read speech

Indian-language speech

Avlys OTS Data

Indian-language speech

Read, extempore and conversational recordings spanning Indian languages, with transcription coverage that varies by release.

Reference collection scope22 languages

Multiple districts and speaking styles

  • Indian-language ASR
  • Speech coverage evaluation
Dataset specifications for Indian-language speech
Formats & labels
Release-specific audio, transcripts and metadata. Transcribed and untranscribed subsets; inspect the chosen version.
Fit & scope
Transcription coverage varies by release. These are human recordings, not human–bot support or escalation calls.
Request a quote by email: Indian-language speech

English audiobook speech

Avlys OTS Data

English audiobook speech

Segmented English audiobook recordings with aligned text, designed for speech-recognition training and evaluation.

Reference collection scopeApproximately 1,000 hours

English · 16 kHz read speech

  • English ASR
  • Transcription benchmarking
Dataset specifications for English audiobook speech
Formats & labels
FLAC audio; text transcripts. Aligned transcripts and clean / other subsets with established splits.
Fit & scope
Narrated books, not spontaneous support calls. Some underlying recordings overlap with the multi-speaker read-speech option.
Request a quote by email: English audiobook speech

Multi-speaker read speech

Avlys OTS Data

Multi-speaker read speech

Sentence-level English recordings with original and normalised text for speech-synthesis research and evaluation.

Reference collection scopeApproximately 585 hours

English · 24 kHz multi-speaker speech

  • Speech synthesis research
  • Text–audio alignment
Dataset specifications for Multi-speaker read speech
Formats & labels
WAV audio; original and normalised text. Sentence-aligned text, speaker information and published splits.
Fit & scope
Read English sentences, not conversational dialogue. Some recordings overlap with the English audiobook-speech option.
Request a quote by email: Multi-speaker read speech

Short spoken commands

Avlys OTS Data

Short spoken commands

Short English words and background noise for keyword-spotting and compact voice-interface experiments.

Reference collection scope105,829 audio files

35 words · 16 kHz · mostly one-second clips

  • Keyword spotting
  • On-device voice recognition
Dataset specifications for Short spoken commands
Formats & labels
WAV audio; labels and split lists. Word labels and speaker identifiers; noise recordings have different lengths.
Fit & scope
Short words and background noise, not dictation or dialogue. The final selection depends on the task.
Request a quote by email: Short spoken commands

Single-speaker read speech

Avlys OTS Data

Single-speaker English speech

Short passages from English books paired with text, recorded by one speaker.

Reference collection scope13,100 clips · about 24 hours

English · 22.05 kHz · single speaker

  • Speech synthesis baselines
  • Audio–text alignment
Dataset specifications for Single-speaker English speech
Formats & labels
16-bit mono WAV; pipe-delimited metadata. Original and normalised text for each clip.
Fit & scope
One speaker only; not suitable for evaluating broad speaker diversity.
Request a quote by email: Single-speaker English speech

Build a useful first batch

What to include
in your brief.

Start with the task your model needs to perform. A scoped pilot establishes the selection, review process and delivery expectations before you scale.

See the delivery checklist
  1. Language, locale, accents and speaking style
  2. Audio hours, speaker coverage and recording conditions
  3. Sample rate, file format and transcript requirements
  4. Speaker-separated evaluation sets and transcript review criteria

Before you choose

Audio data questions.

Pricing, quality & delivery FAQs
Can Avlys help source Hindi and Indian-language speech data?

The catalogue includes Indian-language and multilingual speech options. Specify the language, locale, accent, speaking style and transcript needs so availability can be checked for that scope. A collection's language count alone does not establish that it fits your task.

Is read speech the same as conversational or call-centre data?

No. Read speech, spontaneous conversation and support calls differ in vocabulary, turn-taking, noise and recording conditions. For conversational data, specify the environment, speaker turns, timestamps and redaction requirements. Custom conversation collection and access requirements need a separate scope review.

From selection to scope

Let’s scope your
audio data.

Share the task, coverage, approximate volume and timeline. We’ll discuss availability, sample arrangements and a quote for the agreed work.

Request sample & pricingDiscuss on WhatsApp ↗
sales@avlysai.com

← All Avlys Data services