# Call Recordings

Every voice call is recorded as a dual-channel stereo file. The caller's audio is captured on one channel; the agent's audio is captured on the other. This separation enables independent analysis of each side - post-call quality scoring can evaluate the agent's delivery separately from the caller's speech.

## Recording Format

| Channel              | Format | Sample Rate | Source                       |
| -------------------- | ------ | ----------- | ---------------------------- |
| **Caller (inbound)** | mulaw  | 8 kHz       | Telephony stream             |
| **Agent (outbound)** | PCM16  | 16 kHz      | Text-to-speech engine output |

The two channels are aligned by a shared timeline marker (`media_start_epoch_ms`) that enables precise synchronization during playback and analysis.

## Storage Lifecycle

{% @mermaid/diagram content="flowchart LR
S\[Standard\n0-3 days] -->|3 days| W\[Warm Archive\n3+ days]
W -->|configurable| C\[Cold Archive\nlong-term]
C -.->|restore request| W
W -.->|restore request| S" %}

Recordings move through three storage tiers based on age:

| Tier             | Age       | Access Time               | Use Case                                  |
| ---------------- | --------- | ------------------------- | ----------------------------------------- |
| **Standard**     | 0-3 days  | Immediate                 | Active review, QA, operator playback      |
| **Warm Archive** | 3+ days   | Up to 3 hours to restore  | Compliance review, incident investigation |
| **Cold Archive** | Long-term | Up to 12 hours to restore | Regulatory retention, legal hold          |

When you request a recording that has been archived, the system returns a 202 response with an estimated restoration time. Once restored, the recording is available via a secure, time-limited URL.

## Access

Recordings are accessed through secure, time-limited URLs generated by the Platform API. Each URL is scoped to the requesting workspace and expires after one hour. Metadata includes the call identifier, workspace, direction (inbound/outbound), recording format, sample rate, duration, and byte counts.

{% hint style="info" %}
For how recordings feed into post-call quality scoring, see [Audio Pipeline](/channels/voice/audio-pipeline.md). For how recording data is used in post-call transcript accuracy measurement, see [Performance Characteristics](/reference/performance.md). For the lane-based timeline visualization that synchronizes recordings with agent actions, tool calls, and system events, see [Call Playback Timeline](/intelligence-and-analytics/intelligence.md#call-playback-timeline).
{% endhint %}


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.amigo.ai/channels/voice/recordings.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.