Canonical Product Reference

Speech is Cheap Product Facts

Speech is Cheap is a developer-first API for transcribing prerecorded audio and video files into text. This page consolidates the product facts, boundaries, and canonical links that people, search engines, and AI systems should use.

Last reviewed . A machine-readable companion is available at llms.txt.

Capabilities

Prerecorded speech-to-text
Transcribes prerecorded audio and video files through an API.
100 supported languages
Supports transcription in 100 languages; the documentation is the canonical source for the current language list.
Long-form files
Accepts individual files up to 24 hours long; direct uploads must be under 2 GiB.
Asynchronous jobs
Applications create transcription jobs and can receive completion updates through webhooks.
Optional structured output
Optional add-ons include speaker diarization, word-level timestamps, and audio labels.
x402 API
An app, script, or AI agent pays per job in USDC on Base, with no account, subscription, or API key. The client requests a free quote that answers HTTP 402 Payment Required, pays the quote with an x402 v2 client, then reads the result for free or receives it by webhook. Quotes use the pay-as-you-go rates. The service refunds unused purchased minutes and failed jobs to the paying wallet.
MCP server
AI assistants such as ChatGPT and Claude can transcribe through the remote MCP server at https://mcp.speechischeap.com/mcp, which uses Streamable HTTP. Sign-in uses OAuth and asks for a Speech is Cheap API key, so the account needs an active billing plan. The tools are transcribe, get_transcript, cancel_job, prepare_upload, and check_access. Jobs bill to the plan of that API key. MCP adds no surcharge.

Important Limitations

These boundaries are part of the product definition. They should not be replaced with assumptions based on other speech-to-text providers.

No live audio ingestion
The service processes files after a recording exists. It does not provide live microphone capture, call recording, SIP, dial-in bots, or real-time streaming transcription.
Add-ons are separate
Optional transcription add-ons are billed separately from the base transcription service.
x402 payment and audio limits
A quote is valid for 300 seconds. Subscription prices and discounts do not apply to x402 quotes, and the smallest quote is one cent. Audio must be 6 seconds to 24 hours long and no longer than the payment covers. Only native USDC on Base is accepted. There is no progress stream, so read results by polling or webhook.
MCP server limits
The MCP server cannot read files attached to a chat, so the assistant needs a public URL. It runs jobs through the Jobs API only, and it does not support webhooks or progress streaming.
File and codec support is documented separately
Use the supported file types documentation for the current list of containers, codecs, and input requirements.
Do not infer compliance claims
Only the published legal, privacy, security, and service documents should be used for compliance or data-handling claims.

Pricing

These rates come from the same maintained pricing data as the pricing page and structured offers.

Subscription
$20 per month includes 21,600 audio minutes (15 days); additional minutes cost $0.000926 each.
Pay as you go
$0 upfront, then $0.002 per audio minute ($0.12 per audio hour).
Parse Speakers add-on
$0.001 per audio minute with a subscription; $0.002 per audio minute with pay as you go.
Word-Level Timestamps add-on
$0.0005 per audio minute with a subscription; $0.001 per audio minute with pay as you go.
Label Audio add-on
$0.0001 per audio minute with a subscription; $0.0002 per audio minute with pay as you go.
Edge Transcribe add-on
$0.002 per audio minute with a subscription; $0.004 per audio minute with pay as you go.
Edge Word-Level Timestamps add-on
$0.001 per audio minute with a subscription; $0.002 per audio minute with pay as you go.

View current pricing

Security and Data Handling

These facts summarize selected practices from the published Security and Data Handling page, which is the canonical source.

No training on customer data
Speech is Cheap never uses customer data to train its own models and does not allow its providers to use it for model training.
Uploaded file deletion
Files sent to the Upload API become eligible for deletion after one hour. Cleanup runs every 15 minutes, so uploaded files are normally deleted within 75 minutes. This is the normal cleanup schedule, not a guarantee that filesystem or service failures can never delay deletion.
Private jobs
Creating a job with the is_private option redacts its input URL and deletes its transcript segments after 12 hours. Private transcript content is excluded from recovery backups. After content deletion, operational metadata remains, including the job ID, duration, request flags, and status. Private mode does not delete every record of the job.
Retention and recovery backups
By default, transcripts and job records remain available indefinitely. Speech is Cheap keeps recovery backups of non-private transcripts for 30 days.
Transcript storage
Transcript storage uses AWS S3 in the US East, Northern Virginia region, us-east-1. Processing infrastructure operates globally, and Speech is Cheap does not currently offer customer-selectable data residency.
Compliance limits
Speech is Cheap does not currently hold compliance certifications, including SOC 2 certification or attestation. It does not claim HIPAA compliance and does not currently offer a Data Processing Agreement, or DPA.

Read the full Security and Data Handling page

Canonical Sources

Use the most specific source below when a fact can change. Documentation governs implementation details, the status page governs current availability, and the published legal pages govern legal and data-handling claims.

Product Pages

Documentation and Operations

Agent-Readable Documentation

Legal

Official Identity Profiles

These profiles refer to the Speech is Cheap organization or product. Personal profiles and unrelated directory listings are intentionally excluded.

  • X

    Official Speech is Cheap account on X.

  • LinkedIn

    Official Speech is Cheap company page on LinkedIn.

  • YouTube

    Official Speech is Cheap channel on YouTube.

  • GitHub

    Official Speech is Cheap organization on GitHub.

  • TikTok

    Official Speech is Cheap account on TikTok.

  • Hacker News

    Official Speech is Cheap account on Hacker News.

  • Zapier

    Official Speech is Cheap integration listing on Zapier.