Skip to main content
TalkChief’s AI Transcription automatically converts your calls into accurate, speaker-labeled transcripts across Arabic, Hebrew, and English — including conversations that switch between languages mid-call. Every transcript is diarized, meaning each speaker’s words are clearly attributed and time-stamped, so you can follow the flow of any conversation without listening to the full recording. Transcripts are available through the dashboard and the API the moment processing completes.

​
Supported Languages

TalkChief AI Transcription supports the following languages and combinations:
  • Arabic — Modern Standard Arabic and common regional dialects
  • Hebrew — Full support for Hebrew-language calls
  • English — including accented and regional varieties
  • Mixed-language calls — TalkChief automatically detects when a conversation shifts between Arabic and English or Hebrew and English, and transcribes each segment in the appropriate language
Automatic language detection runs at the start of every call, so you do not need to configure a language per queue or per number. The engine identifies the primary language and adjusts mid-call if the conversation switches. Transcripts use diarization — each turn is labeled with a speaker identifier (e.g., Speaker 1, Speaker 2) and an exact timestamp, giving you a readable, structured record of who said what and when.

​
Pricing

AI Transcription is available as a paid add-on for your TalkChief workspace. Overage minutes are billed at $0.025/min with a one-minute minimum per transcription job. After the first minute, billing continues in per-second increments so you are never charged for unused time. If a transcription job fails for any reason, you are not charged for that job.

​
Enabling AI Transcription

1

Navigate to Settings > Add-ons

Open your TalkChief dashboard, click your workspace name in the top navigation, and select Settings. In the left sidebar, choose Add-ons.
2

Enable AI Transcription for your workspace

Locate the AI Transcription card and toggle it on. You will be prompted to confirm the add-on pricing before activation takes effect.
3

Select queues or call types to transcribe

Once enabled, choose which queues, numbers, or call types should be transcribed automatically. You can enable transcription workspace-wide or limit it to specific inbound or outbound flows.
4

Review transcripts in the AI Transcription dashboard

After your first transcribed calls complete, navigate to AI > Transcription in the main menu. Your transcripts appear here, searchable and filterable by date, agent, queue, and language.

​
Accessing Transcripts

You can access your transcripts in three ways:

Transcription Dashboard

Browse, filter, and read all transcripts directly in the TalkChief dashboard under AI > Transcription. Each transcript includes speaker labels, timestamps, and a link to the associated call recording.

Semantic Search

Use the Insights Dashboard to search across your entire call history using natural language. Find calls by topic, keyword, speaker phrase, or outcome without reading every transcript manually.

API Access

Retrieve transcripts programmatically for downstream processing, CRM sync, or custom reporting. See the Transcription API reference for endpoints, authentication, and response schemas.

Workflow Integrations

Transcripts are available as structured data inputs for TalkChief workflow automations, enabling you to trigger actions based on transcript content or diarized output.

​
Transcript Output

Each transcript is delivered as a diarized, time-stamped document with the following structure:
  • Speaker labels — participants are identified as Speaker 1, Speaker 2, etc. (or by agent name when matched to a TalkChief user)
  • Timestamps — every speaker turn carries a start and end time, accurate to the second
  • Language segments — for mixed-language calls, each segment is tagged with the detected language so you can distinguish Arabic, Hebrew, and English turns at a glance
  • Structured format — transcripts are available as JSON via the API, making them easy to parse, store, and route into CRM systems, ticketing tools, or compliance archives
Transcription accuracy depends on audio quality, background noise levels, and the clarity of each speaker’s audio channel. Always have a qualified human review transcripts before using them for compliance records, legal proceedings, performance evaluations, or any other critical business decision.

​
Frequently Asked Questions

Failed jobs are not billed. If a transcript could not be generated — due to an audio quality issue, processing error, or timeout — the job is marked as failed in the dashboard and no charge is applied. You can retry the transcription for any failed job directly from the transcript list.
Yes — TalkChief automatically detects the language spoken on each call, including mixed Arabic/English or Hebrew/English conversations. You do not need to configure a language in advance. The engine identifies language at the segment level, so a single transcript can contain correctly labeled turns in more than one language.