Multilingual Customer Service Audio Recordings

# Freelancing Requirement: Multilingual Call Center Audio Dataset

We are looking for freelancers, data providers, and agencies who can provide **authentic call-center / customer-service audio recordings** for AI training, speech recognition, conversational AI, and research purposes.

### Audio Requirements

We are looking for **real human-to-human call-center conversations**, including:

* Customer–agent conversations
* Inbound and outbound calls
* Customer support interactions
* Sales and service calls
* Helpdesk / support conversations
* IVR-related conversational data, where available
* Two-way natural conversations

### Languages

We require recordings in both **Indian and International languages**.

**Indian Languages:**
Hindi, Bengali, Tamil, Telugu, Marathi, Gujarati, Kannada, Malayalam, Punjabi, Odia, Assamese and other regional Indian languages.

**International Languages:**
Arabic, Nepali, Spanish, French, German, Portuguese, Italian and other international languages.

We are open to datasets covering **one or multiple languages**.

### Audio Channels

Both formats are required:

* **Mono / Single-channel audio**
* **Stereo / Dual-channel audio**
* Separate speaker channels, where available, are highly preferred.

### Dataset Requirements

* Authentic, human-recorded conversations
* Natural two-way communication
* Clear and usable speech
* Multiple speakers preferred
* Different accents, demographics, and speaking styles are valuable
* Long-form as well as short-form calls are acceptable
* Existing large-scale datasets are preferred

### Metadata / Transcripts

Where available, please also share:

* Transcripts
* Speaker information / diarization
* Timestamps
* Language and dialect
* Call duration
* Channel information
* Call type/category
* Other relevant metadata

These are **preferred but not mandatory** if the underlying audio dataset is strong.

### Privacy & Compliance

All recordings must be legally sourced and available for commercial licensing.

Any personally identifiable or confidential information should be appropriately **masked, removed, or de-identified** before delivery.

The provider must have the necessary rights and permissions to license the recordings for **commercial AI/ML training, evaluation, research, and related model development purposes**.

### Please Include in Your Proposal

1. Languages available
2. Total recording hours
3. Number of calls/files
4. Mono vs. stereo availability
5. Separate speaker-channel availability
6. Average call duration
7. Transcript/metadata availability
8. Data source and collection methodology
9. Privacy/de-identification process
10. Licensing and commercial usage rights
11. Sample audio for evaluation
12. Pricing and delivery timeline

**We are interested in both individual language datasets and large multilingual collections. High-volume datasets are especially welcome.**