Moving from Hume AI Octave to ReadAloud

Last verified October 10, 2026.

Details about Hume AI Octave come from its public pages as they were on October 10, 2026. Hume AI Octave may have changed them since, and a page that was unclear to us may be clear to you. Corrections: support@readaloudai.org. We make ReadAloud, so read this page with that in mind and check anything important with Hume AI Octave directly.

Moving from Hume AI Octave: what kind of move this is

How Hume AI Octave positions itself: Hume's Octave is a speech-language-model text-to-speech API with prompt-based voice design and acting instructions; Hume's documentation says its TTS and EVI APIs are being sunset, with access ending on 13 November 2026.

Hume AI Octave is a model vendor, and speech is one endpoint among its other services. Moving speech alone leaves the rest of your integration in place.

Hume AI Octave's pages suggest it fits: teams that were using Octave for expressive, prompt-designed voices; because Hume has announced the API's shutdown on 13 November 2026, new projects should plan around that date.

Before you leave Hume AI Octave

• Export what you need from your Hume AI Octave account before November 13, 2026. Its documentation says account data is deleted afterwards, so audio, voice designs and settings you have not saved will be gone. (source: dev.hume.ai)
• Collect the exact text you send to Hume AI Octave today: the longest, the shortest, and the ones with names, numbers and abbreviations.
• Conditions on Hume AI Octave's side: Users retain rights to output; Hume receives a perpetual licence to use voice recordings and voice models to provide and improve services. Use must comply with Hume's prohibited-use policy. Check what they mean for audio you already generated. (source: dev.hume.ai)
• Commercial use at Hume AI Octave: Commercial use was allowed on Creator and above; Free and Starter were for non-commercial use. (source: dev.hume.ai)
• List where your code uses TypeScript, Python, .NET or CLI so you can replace each call site.
• Condition listed by Hume AI Octave: Free and Starter plans were limited to non-commercial use. (source: dev.hume.ai)
• Save any audio you need to keep from Hume AI Octave before you cancel, and note which plan covers the right to keep using it.

Steps to move off Hume AI Octave

These steps come from Hume AI Octave's public pages. Sources are listed at the end.

1. Plan the move before the 13 November 2026 shutdown date stated by Hume; account data is deleted afterwards, so export anything you need first. (source: dev.hume.ai)
2. If you also use Hume's EVI speech-to-speech API, it ends on the same date; this checklist covers the text-to-speech API, not EVI. (source: dev.hume.ai)
3. Hume authenticates with an API key (the SDK reads HUME_API_KEY); replace the SDK client with the target service's client. (source: dev.hume.ai)
4. Replace /v0/tts, /v0/tts/file or the streaming endpoints (/v0/tts/stream/json, /v0/tts/stream/file) with the target's equivalents; WebSocket users replace /v0/tts/stream/input. (source: dev.hume.ai)
5. Map each utterance (text, voice name or id, description, speed, trailing_silence) to the target's request fields; description-based voice design and acting instructions have no direct equivalent in most services. (source: dev.hume.ai)
6. Re-create cloned voices on the target from the original consented audio, and re-test output format needs (MP3, WAV or PCM). (source: dev.hume.ai)

Map each Hume AI Octave request field and endpoint

What changes in your code and what does not carry over

• Endpoint: Hume AI Octave's request shape is its own, so rewrite the call; /v1/audio/speech is one plain HTTP request.
• Voice: Hume AI Octave voices do not exist at ReadAloud. Live: one American English voice. Studio: several English voices.
• Languages, as Hume AI Octave's pages put it: Octave 1: English and Spanish. Octave 2 (preview): English, Japanese, Korean, Spanish, French, Portuguese, Italian, German, Russian, Hindi, Arabic. ReadAloud: English today. (source: dev.hume.ai)
• Formats: Hume AI Octave lists MP3, WAV and PCM. ReadAloud returns mp3, wav and pcm on at least one route.
• Request size: Hume AI Octave states 5,000 characters per utterance; description up to 1,000 characters; up to 5 generations per request. ReadAloud accepts 5,000 characters per request, so split longer text at sentence boundaries. (source: dev.hume.ai)
• Streaming: Hume AI Octave lists WebSocket streaming. ReadAloud streams over WebSocket after an authorize call; see /developers.
• Custom voices, as Hume AI Octave's pages describe them: Voice design from a text description; voice cloning from as little as 15 seconds of audio (availability depends on subscription tier). Those voices cannot be exported to ReadAloud. (source: dev.hume.ai)

Where ReadAloud does not match Hume AI Octave

• Expressive control: Hume AI Octave's pages list controls over emotion, acting or style. ReadAloud does not offer emotion or expressive control: it has no acting instructions, no style prompts, no SSML and no audio tags, and the instructions field of the OpenAI route is ignored. If you need that, ReadAloud is not the right fit for that part of your product; evaluate other vendors that list it. (source: dev.hume.ai)

Steps on the ReadAloud side

1. Set the cut-over date before November 13, 2026, and keep Hume AI Octave running until the ReadAloud path has carried real traffic.
2. Create a key (/developers#get-started) and run the test call below.
3. Pick ReadAloud Live (the low-latency tier, built for live calls and voice agents) or ReadAloud Studio (the higher-priced tier for read-aloud and narration, with several English voices).
4. Keep the base URL, key and voice in a setting, and move a small share of traffic first.

# ReadAloud through its OpenAI-compatible speech route (/v1/audio/speech).
# Replace YOUR_KEY with a ReadAloud API key.
curl -X POST "https://api.readaloudai.org/v1/audio/speech" \
  -H "Authorization: Bearer YOUR_KEY" -H "Content-Type: application/json" \
  -d '{"model":"tts-1","voice":"readaloud-default","input":"Paste a sentence you send to Hume AI Octave today.","response_format":"mp3"}' \
  -o test.mp3

What it costs to test ReadAloud

A one-time grant of free credits worth $0.10 (about 10,000 characters of speech), shared by all keys on the account.

Hume AI Octave no longer has a pricing page on its own site; the copy we found is an archived snapshot. We show no Hume AI Octave price, because a price that cannot be checked against the vendor's own page may be out of date.

What we could not confirm about Hume AI Octave

• Whether Hume will change the 13 November 2026 date or offer an export route for voices and data; its documentation states that access ends and data is deleted afterwards, with no further detail.
• The live pricing page no longer exists (it redirects to the homepage); plan figures come from an Internet Archive snapshot dated 2026-09-01, so no price is shown.
• Total voice count is not stated.
• SSML support: docs describe acting instructions and no SSML page was found.
• Whether pricing changed between the 2026-09-01 snapshot and the sunset announcement.
• The concurrent-connections row on the archived page sits under the speech-to-speech (EVI) section, so TTS concurrency is not stated.
• Whether cloned voices can be exported before the shutdown.

Sources

• Hume docs - welcome (sunset notice) (retrieved October 10, 2026)
• Hume docs - Text-to-Speech overview (retrieved October 10, 2026)
• Hume docs - TTS FAQ (retrieved October 10, 2026)
• Hume docs - billing (retrieved October 10, 2026)
• Hume docs - TTS Python quickstart (retrieved October 10, 2026)
• Hume AI pricing page (Internet Archive snapshot, 2026-09-01) (retrieved October 10, 2026)