As of November 2026, Hume's own product list reads Audio Decisions API, Expression APIs, Human Feedback API, User Testing API, and Speech Data, with no text to speech on it. The old Octave and EVI documentation addresses now redirect (HTTP 308) to the Audio Decisions docs, Hume's status page tracks two APIs, and the product pages for Octave and EVI return 404, while the Octave 2 launch post dated 1 October 2025 is still online.
A team that built a voice on Octave therefore needs somewhere to send its scripts. Bland Speech leads the five options here, since Bland Speech v3 ranked first of 13 AI voices in Design Arena's Audio Realism Bench results of 4 August 2026 and a Bland Speech script carries its own delivery as tags. ElevenLabs, Fish Audio, Google Cloud, and Cartesia are the other four, and every provider outside them has a place on the text to speech hub or the alternatives family page.
Why do teams look for Hume Octave alternatives?
Teams look for Hume Octave alternatives because Hume's site no longer shows a text to speech product, and Octave's draw was delivery that followed the emotion in a script.
Octave 2, announced on 1 October 2025, was described by Hume as a speech-language model that "more deeply understands the emotional tone of speech". Hume said it spoke 11 languages, generated audio in under 200ms, cloned its sample voices from 15-second recordings, and added voice conversion and phoneme editing. Neither that post nor the EVI 3 post of 17 July 2025 carries a notice that the product ends.
What a team replaces is a list of jobs rather than one feature: direct the emotion of a line, clone a voice from seconds of audio, and stay fast enough for a live call. Hume's current homepage leads with a different business: "Measure and improve real-world voice AI performance".
What should I check in a Hume Octave alternative?
A buyer should check how delivery is directed, how much audio a clone needs and how the speaker is verified, what phone audio comes out, and how each character is billed.
Direction comes in three forms. Tags are words in square brackets placed before the speech they shape: Bland Speech takes [laughs] and [say warmly], ElevenLabs takes [whispering] and [shouting] on Eleven v4, and Fish Audio takes free-form brackets such as [whispers sweetly]. A style prompt describes the delivery in plain language, which Google's Gemini TTS uses beside its tags. Numeric controls turn a dial: Bland Speech has expressiveness and stability from 0.0 to 1.0, and Cartesia has speed and volume plus emotion controls that are Beta and English only.
Cloning differs in audio and gatekeeping. Instant clones need about 10 seconds of audio on Bland Speech, ElevenLabs' Eleven v4, Cartesia, and Fish Audio, while Google's instant custom voice is closed to new requests. ElevenLabs verifies a professional clone with a voice captcha, and Bland puts a professional clone through a consent and rights review.
What are the best Hume Octave alternatives?
The best Hume Octave alternatives are Bland Speech, ElevenLabs, Fish Audio, Google Cloud, and Cartesia, with Bland Speech first for phone calls and voice agents because of its realism result, inline tags with two numeric controls, and one standard per-character rate.
What these rows mean
- Price per million characters: each company's published rate with the model or plan named. Fish Audio bills per million UTF-8 bytes, not characters, and charges $50 for drama-3, with a $10 launch price through 30 November 2026.
- Audio Realism Bench: Design Arena's blind listening test, with rank among 13 AI voices and an Elo score. The scored model is named, and Google Cloud's row is Gemini TTS, not Chirp 3: HD.
1. Bland Speech
Bland Speech is a text to speech model that Bland runs itself, with a studio to try voices in and an API to build on. A script directs its own delivery: tags such as [say warmly], [say angrily], [sighs], and [clears throat] sit right before the words they shape, a pause is written as <|0.5|> for half a second, and two numeric controls, expressiveness (how much intonation varies) and stability (how much the same input varies between renders), run from 0.0 to 1.0. In the studio, a director feature called Norm writes the words, the stumbles, and the timing from a scene description, and performs them in a single speaker's voice.
The model suits a team moving an Octave voice onto phone calls. An instant clone needs one clean sample of about 10 seconds, a professional clone is fine-tuned on 30 minutes or more of audio after Bland's consent and rights review, and every clone is private to the customer's organization. Output is PCM from 8 to 48 kHz or mu-law at 8 kHz, and the text to speech API page has the endpoints.
Pros
- Bland Speech v3 ranked first of 13 AI voices in Audio Realism Bench results of 4 August 2026
- Inline tags, timed pauses, and two numeric controls in the script itself
- An instant clone from about 10 seconds of audio, private to the customer's organization
- Mu-law at 8 kHz and PCM from 8 to 48 kHz for phone lines
Cons
- More than a hobby project needs, because Bland Speech is built for production calls
Pricing: $0.04 per 1,000 characters ($40 per million), pay as you go with no monthly plan, and 50,000 free characters once for a new account. Enterprise pricing is custom, set with Bland's sales team. The Bland Speech page has the studio, and how to migrate from Hume Octave has the steps.
2. ElevenLabs
ElevenLabs describes Eleven v4, launched on 28 September 2026, as its "most emotionally rich, expressive speech synthesis model", and the v4 family speaks 90+ languages. The API suits a team that wants audio tags and a plan ladder: an instant clone from 10 seconds starts on Starter at $6 a month, and a professional clone starts on Creator at $22 and is verified with a voice captcha.
Pros
- Audio tags such as [whispering], [shouting], and [laughing] on Eleven v4
- 90+ languages across the Eleven v4 family
- Professional clones verified with a voice captcha
Cons
- Eleven v4 and Eleven v3 don't support SSML break tags, so pauses come from punctuation and tags
- Eleven v4 bills $0.08 per 1,000 characters, twice Eleven v4 Turbo's $0.04
- The Free plan is non-commercial, and a commercial licence starts on Starter
Pricing: $0.08 per 1,000 characters for Eleven v4 and $0.04 for Eleven v4 Turbo and Flash, billed in dollars on the API. Plans run Starter $6 a month, Creator $22, Pro $99, Scale $299, and Business $990, with Enterprise custom.
3. Fish Audio
Fish Audio launched Drama 3 on 13 October 2026, a model it built for voice direction, where plain-language instructions add a sound or a pause where it happens or limit a delivery to specific words. Fish Audio's S2 models take bracket tags that aren't limited to a fixed set, such as [whispers sweetly] and [laughing nervously], and Fish Audio claims 15,000+ direction tags. The API suits a team that wants to pay for usage with no subscription and direct delivery in its own words, and Bland Speech vs Fish Audio sets the two side by side.
Pros
- Free-form bracket tags instead of a fixed list
- No subscription or monthly minimum for API access
- An instant clone from a clip of at least 10 seconds
Cons
- TTS output formats are mp3, wav, pcm, and opus, so a team converts to mu-law itself
- A SOC 2 audit is underway, with the report to follow under NDA once complete
- Drama 3 costs $50 per million UTF-8 bytes once a $10 launch price ends on 30 November 2026
Pricing: The API bills per million UTF-8 bytes: $15.00 for S2.1-Pro and $50.00 for drama-3, with a free S2.1-Pro tier under fair-use limits. Fish Audio converts 1 million bytes to about 180,000 English words.
4. Google Cloud
Google Cloud Text-to-Speech sells Chirp 3: HD voices alongside Gemini TTS models priced per token. Gemini TTS takes four groups of inline tags, for sounds such as [sigh], style such as [whispering], vocalized adjectives such as [curious], and pauses such as [short pause], and a style prompt sets the overall tone. Google warns that vocalized-adjective tags are spoken aloud, "likely an undesired side effect for most use cases". Chirp 3: HD has no emotion control, only a speaking rate from 0.25 to 2.0 and pause tags.
Pros
- Gemini 3.1 Flash TTS offers single-speaker and multi-speaker dialogue
- Text-to-Speech sits inside Google Cloud's HIPAA BAA and SOC 2 scope
- The first million Chirp 3: HD characters are free each month
Cons
- Chirp 3: Instant Custom Voice is closed to new access requests, and Gemini 3.8 voice replication is in Preview
- Gemini voices bill per token, so a per-character budget needs a conversion
Pricing: Chirp 3: HD costs $30 per million characters after a free first million each month. Gemini 2.5 Pro TTS costs $1.00 per million text input tokens and $20.00 per million audio output tokens, at 25 audio tokens per second of speech.
5. Cartesia
Cartesia sells Sonic 3.6, a model that speaks 44 languages, with a current snapshot dated 27 August 2026. Cartesia says Sonic delivers first audio in under 90ms. Cartesia's emotion controls suit a team that writes English scripts and accepts a Beta feature: Cartesia says emotion tags are supported only for English, and a professional voice clone fixes speed and volume at what the training data taught it.
Pros
- Speed from 0.6 to 1.5 and volume from 0.5 to 2.0 on a generation
- An instant clone from 10 seconds, with up to 60 seconds to hold an accent better
- Mu-law and A-law output at 8 kHz
Cons
- Emotion controls are Beta and English only
- Instant cloning needs the Pro plan, and professional cloning needs Startup and 30 minutes of audio
- Credits, rollover limits, and overage settings to manage each month
Pricing: Free is $0 a month with 20,000 credits, Pro $5 with 100,000, Startup $49 with 1.25 million, and Scale $299 with 8 million, with Enterprise custom. Overages cost $65, $45, and $38 per million credits on Pro, Startup, and Scale.
Which Hume Octave alternative sounds most human on a call?
Bland Speech v3 ranked first of 13 AI voices at 1365 in Design Arena's Audio Realism Bench results of 4 August 2026, and the human recording scored 1500. Listeners hear two unlabelled recordings of one line and pick the more human one, and Elo turns those picks into a score.
- #1BlandBland Speech v31365
- 5thGoogleGemini 2.5 Pro TTS1109
- 6thCartesiaSonic 3.51096
- 12thElevenLabsEleven v3Near the bottom938
Ranks are among 13 AI voices, and each rival shows the highest listed voice available on it. Real human recordings scored 1500, at the right edge of this scale. Bars start at 600, not zero. Source: Design Arena's published results.
Of the other options, Google's Gemini 2.5 Pro TTS ranked 5th at 1109, Cartesia's Sonic 3.5 ranked 6th at 1096, and ElevenLabs' Eleven v3 ranked 12th at 938. Hume and Fish Audio weren't in those results, and Cartesia's Sonic 3.6 (27 August 2026) and ElevenLabs' Eleven v4 (28 September 2026) came after them and weren't scored in them.
The benchmark measures conversational realism, with phone-agent prompts among its categories, and Design Arena sets acting-heavy speech such as game characters and dramatized fiction out of scope. A team that wants a performed character rather than a caller-facing agent should audition Fish Audio's Drama 3 and ElevenLabs' Eleven v4 before settling.
Which Hume Octave alternative should I pick?
For a phone agent or a voice agent, Bland Speech is the pick: Bland Speech v3 ranked first of 13 AI voices in those 4 August 2026 results, and delivery lives in the script as tags, pauses, and two numeric controls.
Use Bland Speech when
- A team will put an expressive voice on phone calls and write delivery into each script
- A team will clone a voice from about 10 seconds of audio and confirm the voice is its to use
Use another option when
- A team will direct performances in free-form plain language: Fish Audio
- A team will manage plan tiers for 90+ languages and audio tags: ElevenLabs
- A team already runs on Google Cloud and wants style prompts on Gemini voices: Google Cloud
- A team will manage monthly credit allowances and write English scripts with emotion controls: Cartesia
On price, Google's Chirp 3: HD at $30 per million characters and Fish Audio's S2.1-Pro at $15 per million UTF-8 bytes both undercut Bland Speech's $40, so a script-heavy workload that doesn't need Bland Speech's phone realism should price them first.
Frequently asked questions
What happened to Hume Octave?
Hume's own site no longer shows a text to speech product. The old Octave and EVI documentation addresses redirect to the Audio Decisions docs, and Hume's status page tracks two APIs. The Octave 2 launch post of 1 October 2025 is still online.
What is the best Hume Octave alternative for phone calls?
Bland Speech is the best Hume Octave alternative for phone calls, because Bland Speech v3 ranked first of 13 AI voices in Design Arena's results of 4 August 2026, it returns mu-law at 8 kHz, and a script carries its own delivery in tags. Bland publishes 199ms to first audio, against the under 200ms Hume claimed for Octave 2.
Which Hume Octave alternatives take emotion tags in the script?
All five do in some form. Bland Speech takes [say warmly] and [laughs], ElevenLabs takes [whispering] on Eleven v4, Fish Audio takes free-form brackets, Google's Gemini TTS takes sound, style, adjective, and pause tags, and Cartesia takes emotion tags in English only, as a Beta feature.
Can an Octave voice move to another provider?
No, a cloned voice stays with the vendor that built it, so a team re-clones from the original recordings with consent. Hume cloned its Octave 2 sample voices from 15-second recordings, and Bland Speech needs one clean sample of about 10 seconds and the customer's confirmation that the voice is theirs to use.
Which Hume Octave alternative costs the least?
Google Cloud's Chirp 3: HD is $30 per million characters with a free first million each month, and Fish Audio's S2.1-Pro is $15 per million UTF-8 bytes. Bland Speech is $40 per million characters with 50,000 free characters once, and ElevenLabs runs $40 to $80 by model.
Does Bland Speech have emotion controls like Octave?
Bland Speech has three: bracket tags such as [say warmly] and [sighs], timed pauses written as <|0.5|>, and expressiveness and stability controls from 0.0 to 1.0.
