Dataset Viewer
Auto-converted to Parquet Duplicate
call_id
stringclasses
4 values
caller_id
stringclasses
4 values
turn
int64
1
7
speaker
stringclasses
2 values
start_s
float64
0
33
end_s
float64
3
39.3
language
stringclasses
2 values
text
stringlengths
10
92
intent
stringclasses
4 values
sentiment
stringclasses
3 values
audio_file
stringclasses
4 values
consent_ref
stringclasses
4 values
CALL-2001
clr_663c5b12
1
agent
0
4.3
en
Thank you for calling, how can I help you today?
delivery_delay
negative
call_2001.flac
src-0221 / v1
CALL-2001
clr_663c5b12
2
customer
4.7
11.8
hi-en
Haan hi, mera order [REDACTED] abhi tak deliver nahi hua, it was supposed to come yesterday.
delivery_delay
negative
call_2001.flac
src-0221 / v1
CALL-2001
clr_663c5b12
3
agent
12.7
18.1
en
Sorry about that. Can you confirm the phone number on the order?
delivery_delay
negative
call_2001.flac
src-0221 / v1
CALL-2001
clr_663c5b12
4
customer
18.6
22
en
Yes, it's [REDACTED].
delivery_delay
negative
call_2001.flac
src-0221 / v1
CALL-2001
clr_663c5b12
5
agent
22.6
29
en
Thanks. I can see it's at the local hub and should reach you by tomorrow evening.
delivery_delay
negative
call_2001.flac
src-0221 / v1
CALL-2001
clr_663c5b12
6
customer
29.9
36.3
hi-en
Theek hai, but please make sure it comes tomorrow, I need it urgently.
delivery_delay
negative
call_2001.flac
src-0221 / v1
CALL-2002
clr_e0927a9d
1
agent
0
4.1
en
Good afternoon, you're speaking with support.
refund_request
neutral
call_2002.flac
src-0222 / v1
CALL-2002
clr_e0927a9d
2
customer
4.6
11.2
en
Hi, I returned a pair of shoes last week and I still haven't got the refund.
refund_request
neutral
call_2002.flac
src-0222 / v1
CALL-2002
clr_e0927a9d
3
agent
11.7
15.6
en
I can check that. What's the return number?
refund_request
neutral
call_2002.flac
src-0222 / v1
CALL-2002
clr_e0927a9d
4
customer
16
21.1
en
It's [REDACTED]. They picked it up on Monday.
refund_request
neutral
call_2002.flac
src-0222 / v1
CALL-2002
clr_e0927a9d
5
agent
22
28.9
en
The pickup is confirmed. The refund was issued today and takes 3 to 5 working days to show.
refund_request
neutral
call_2002.flac
src-0222 / v1
CALL-2002
clr_e0927a9d
6
customer
29.3
32.9
en
Okay, thanks for checking.
refund_request
neutral
call_2002.flac
src-0222 / v1
CALL-2003
clr_95ea784c
1
agent
0
3
en
Hello, how may I help?
billing_dispute
negative
call_2003.flac
src-0223 / v1
CALL-2003
clr_95ea784c
2
customer
3.6
11.4
hi-en
Mujhe is month ka bill zyada aaya hai, I think I've been charged twice for the same plan.
billing_dispute
negative
call_2003.flac
src-0223 / v1
CALL-2003
clr_95ea784c
3
agent
12.3
18.4
en
Let me look at your account. Could you share the account number?
billing_dispute
negative
call_2003.flac
src-0223 / v1
CALL-2003
clr_95ea784c
4
customer
19.2
21.3
en
[REDACTED]
billing_dispute
negative
call_2003.flac
src-0223 / v1
CALL-2003
clr_95ea784c
5
agent
21.6
28.8
en
I can see two charges on the 3rd. One is a duplicate, I'll raise a reversal now.
billing_dispute
negative
call_2003.flac
src-0223 / v1
CALL-2003
clr_95ea784c
6
customer
29.2
32.7
hi-en
Okay good, kitne din lagenge?
billing_dispute
negative
call_2003.flac
src-0223 / v1
CALL-2003
clr_95ea784c
7
agent
33
39.3
en
It should reflect within 7 days. You'll get an SMS confirmation.
billing_dispute
negative
call_2003.flac
src-0223 / v1
CALL-2004
clr_b18744d7
1
agent
0
3.8
en
Clinic front desk, how can I help?
appointment_booking
positive
call_2004.flac
src-0224 / v1
CALL-2004
clr_b18744d7
2
customer
4.5
10.5
en
I'd like to book an appointment with the skin doctor this week.
appointment_booking
positive
call_2004.flac
src-0224 / v1
CALL-2004
clr_b18744d7
3
agent
10.8
15.4
en
We have Thursday at 11 or Friday at 4. Which works?
appointment_booking
positive
call_2004.flac
src-0224 / v1
CALL-2004
clr_b18744d7
4
customer
16.1
19.4
hi-en
Friday 4 baje theek rahega.
appointment_booking
positive
call_2004.flac
src-0224 / v1
CALL-2004
clr_b18744d7
5
agent
19.7
24.4
en
Done. Can I have the patient's name and number?
appointment_booking
positive
call_2004.flac
src-0224 / v1
CALL-2004
clr_b18744d7
6
customer
24.7
29.6
en
[REDACTED], and the number is [REDACTED].
appointment_booking
positive
call_2004.flac
src-0224 / v1

Call-centre transcripts: sample

Turn-level transcripts of customer support calls, with speaker, timestamps, a language tag per turn (English or Hindi-English code-mixed), intent and sentiment. This file shows the schema and labels of the call-centre speech data that Datadhara licenses to AI teams.

What this is

  • 25 rows: 4 calls, 6 to 7 turns each, one row per speaker turn.
  • Calls cover four intents: delivery_delay, refund_request, billing_dispute, appointment_booking.
  • 5 of the 25 turns are code-mixed Hindi-English (hi-en), switching language mid-sentence. The rest are English.
  • Phone numbers, order, return and account numbers, and names are replaced with [REDACTED]. Callers are identified by a pseudonymous ID.

The rows in this file were created to show the format, labels and de-identification. They do not come from real calls, and the consent_ref values are format examples. Licensed deliveries contain real recordings collected under written agreements.

Fields

Column Type Description
call_id string Call identifier, e.g. CALL-2001
caller_id string Pseudonymous caller ID, e.g. clr_663c5b12
turn int Turn number within the call, starting at 1
speaker string agent or customer
start_s float Turn start time in seconds from the start of the call
end_s float Turn end time in seconds
language string en or hi-en (Hindi-English code-mixed)
text string Transcript of the turn, with identifiers replaced by [REDACTED]
intent string Call intent (call-level label, repeated on each turn)
sentiment string positive, neutral or negative (call-level label, repeated on each turn)
audio_file string Name of the matching audio file, e.g. call_2001.flac (audio not included in this sample)
consent_ref string Reference to the source agreement and its version, e.g. src-0221 / v1

Example row

{
  "call_id": "CALL-2001",
  "caller_id": "clr_663c5b12",
  "turn": 2,
  "speaker": "customer",
  "start_s": 4.7,
  "end_s": 11.8,
  "language": "hi-en",
  "text": "Haan hi, mera order [REDACTED] abhi tak deliver nahi hua, it was supposed to come yesterday.",
  "intent": "delivery_delay",
  "sentiment": "negative",
  "audio_file": "call_2001.flac",
  "consent_ref": "src-0221 / v1"
}

Load it

from datasets import load_dataset

ds = load_dataset("datadhara/call-centre-transcripts-sample", split="train")

Intended uses

Use this sample to check the schema and label set against your pipeline before requesting real data. The licensed data is used for:

  • Speech recognition (ASR) for accented, noisy and code-mixed speech
  • Intent detection and call summarisation
  • Evaluation sets for voice agents

Limitations of this sample

  • Text only. The audio_file column names the audio a delivery would include; no audio is in this repository.
  • Four calls is enough to see the format, not to train or benchmark anything.
  • intent and sentiment are call-level labels repeated on every turn.

Full dataset and licensing

This sample file is released under CC BY 4.0, as on the samples page.

The full data is not on Hugging Face. It is available under a commercial licence. Datadhara collects to your spec: support calls and sales calls as WAV or FLAC with JSON transcripts, speaker turns, timestamps and intent or outcome labels. Languages, volume and labels are agreed per project. Licences cover training and evaluating commercial models, and can be exclusive or non-exclusive.

To request data, describe what you need at datadhara.in/speech-data-for-ai or email hello@datadhara.in.

Consent and de-identification

As described in Datadhara's methodology:

  • Every source organisation signs a written agreement that explicitly covers use for AI training. Audio is recorded only where participants have been told and agreed.
  • Names, phone numbers, order and account numbers are masked in both the transcript and the audio. Calls that mention sensitive personal details beyond what the spec needs are dropped.
  • Transcripts are drafted automatically, then checked and corrected by people.
  • Every record carries a consent_ref linking it to its source agreement. In a real delivery, that reference resolves to the signed agreement.

Citation

The same sample files are archived on Zenodo: https://doi.org/10.5281/zenodo.24349905

Downloads last month
-