> ## Content Index
> Fetch the complete content index at: https://www.theaidirectory.com.au/llms.txt
> Use this file to discover other available public pages before exploring further.

# Fish Audio Review (Australia 2026): Cheap, Fast Voice Cloning — With Serious Consent Caveats
- URL: https://www.theaidirectory.com.au/review-fish-audio-australia-2026/
- Published: 2026-06-25T23:00:00.000Z
- Updated: 2026-10-09T06:30:04.000Z
- Description: Fish Audio offers text-to-speech and voice cloning at some of the lowest prices in the market. The S1 model is impressively expressive. But the low barrier to cloning — and the community voice library — carries real consent and legal risks Australians need to understand.
- Author: Adam
- Tags: Audio, Voice & Podcasting, AI Reviews, #review

**13**/25

★★★☆☆

Good

Fish Audio · name: Fish Audio

Verified 1 Oct 2026 Reviewed 26 Jun 2026 

Value for Money**4**/5

Ease of Use**3**/5

Feature Depth**3**/5

Australian Fit**1**/5

Reliability & Support**2**/5

From

**A$7.70**/mo · ex-GST · billed in USD ($5.50)

**The gate:** Five stars need 21+ and Australian Fit of at least 4/5.

[Skip to the verdict ](#verdict) 

*Pricing figures updated 6 September 2026: USD prices are converted at the current rate (1 USD ≈ 1.40 AUD, verified 5 September 2026) and shown ex-GST, replacing the retired 1.55 + GST basis. A GST-registered Australian business that supplies its ABN is not charged GST on an imported digital service; other buyers should allow \~10% on top (*[*Methodology*](https://www.theaidirectory.com.au/methodology/)*).*

> **Verdict (TL;DR):** Fish Audio is a low-cost text-to-speech and voice cloning platform with a capable S1 model, a massive community voice library, and API access from the first dollar. For developers and creators who need affordable TTS, it delivers. The voice cloning feature is impressive technically — 10 seconds of audio is enough to generate a replica. That low barrier is precisely the problem: it makes cloning someone else's voice without consent dangerously easy. Australian users should be clear on the consent requirements under privacy and intellectual property law before using this feature commercially. Data also goes offshore.  
>  
> **Score: 13/25 — ⭐⭐⭐ Good** *Last reviewed: 26 June 2026 by Adam.*

---

## What Fish Audio is

Fish Audio is a text-to-speech and voice cloning platform founded in 2024 by Rissa Cao, Shijia Liao, and Jiahua Liu. The company is incorporated in the US (Mountain View, California) and operates a cloud API alongside a web interface.

The headline features are:

- **Text-to-speech** using the S1 model — Fish Audio's highest-quality voice generation model, trained with reinforcement learning from human feedback (RLHF) and capable of over 48 emotion expressions.
- **Voice cloning** using just 10 seconds of reference audio — lower than most competitors that require 60 seconds or more.
- **Community voice library** — over 2 million community-created voice models, publicly searchable and usable via API.
- **Open-source foundation** — Fish Speech (the underlying model) is open-source on GitHub, which enables self-hosted deployment for teams with the technical capability.

Fish Audio competes on price and accessibility. It is not trying to be ElevenLabs. The target user is a developer or creator who needs good-enough TTS at the lowest possible cost, or who wants to clone their own voice for a specific production use case.

---

## What it costs in Australia

Fish Audio prices in USD. All figures below are approximate AUD conversions at \~1.40 USD/AUD. Shown ex-gst (a gst-registered business quoting its abn is not charged gst on an imported digital service; other buyers add \~10%)..

### Subscription plans

| Plan | USD/month               | Approx AUD/month (ex-GST) | Notes                                                            |
| ---- | ----------------------- | ------------------------- | ---------------------------------------------------------------- |
| Free | $0                      | $0                        | 8,000 credits; personal use only; no commercial rights           |
| Plus | \~$5.50/month (annual)  | \~AUD $7.70               | 250,000 credits; \~200 min/month S1; commercial use; API access  |
| Pro  | \~$37.50/month (annual) | \~AUD $52.50              | 2,000,000 credits; \~27 hrs/month S1; commercial use; API access |

*Promotional pricing applies as of mid-2026\. Standard (monthly billing) list prices are higher — Plus at $20/month, Pro at $150/month. Confirm current pricing at fish.audio/plan.*

### API pricing (pay-as-you-go)

| Model           | USD per million UTF-8 bytes | Notes                                                    |
| --------------- | --------------------------- | -------------------------------------------------------- |
| S1 / all models | USD $15/million bytes       | \~AUD $21.00/million bytes ex-GST; no minimum commitment |

One million UTF-8 bytes is approximately 180,000 English words or roughly 12 hours of audio. For context, a typical 500-word blog post converted to speech costs a fraction of a cent.

---

## How it works in practice

Fish Audio has both a web interface and an API. The web interface allows you to type or paste text, select a voice (from the community library or a cloned voice you've created), and generate audio. This is accessible to non-technical users for personal or light commercial use.

The API follows standard patterns. You authenticate with an API key, POST a request with your text and a voice reference ID, and receive audio back as a stream or file. The Python SDK and direct HTTP access are both documented.

**Voice cloning** is the standout feature technically. Upload 10–30 seconds of clear audio and Fish Audio generates a voice model you can use for TTS. Quality depends heavily on the reference audio — clean, studio-quality recordings produce much better results than phone-quality audio. The S1 model can capture emotion and speaking style from the reference.

**The community voice library** contains over 2 million voices, searchable by language, gender, and style. Many are genuinely useful. Some raise questions about consent: it is not always clear whether a community-uploaded voice was created with the permission of the person being cloned.

Performance is fast. S1 generation for a typical sentence takes 1–3 seconds via API.

---

## What it does well

**Price is hard to beat.** At USD $15 per million bytes for API usage, Fish Audio is among the cheapest production-ready TTS APIs available. For high-volume, cost-sensitive applications, this matters.

**S1 model quality is genuinely good.** The RLHF-trained S1 model supports a wide range of emotions and produces natural-sounding speech that is competitive with mid-tier offerings from larger providers. It ranked highly on independent TTS Arena benchmarks in 2025–2026.

**Voice cloning threshold is low.** Ten seconds of audio is all that's required. For cloning your own voice, this is excellent. It means you don't need a recording studio session to create a usable voice model.

**Open-source option exists.** Fish Speech (the underlying model) is available on GitHub. Technical teams with the capacity to self-host can run it locally — which addresses data residency concerns and eliminates the per-request cost.

**API access from the first dollar.** Unlike some platforms that gate API access behind expensive plans, Fish Audio's API is available on pay-as-you-go from the start.

---

## What it doesn't do well

**Voice cloning consent is the platform's biggest risk — and it is largely offloaded to the user.** Fish Audio's terms require users to confirm they have the rights and consents required for any voice they clone. In practice, the 10-second sample requirement means it is trivially easy to clone someone's voice from a publicly available recording. The community voice library contains models where consent provenance is unclear.

Under Australian law, cloning a real person's voice without their consent and using it commercially carries serious risk. The Privacy Act 1988 covers biometric data. Intellectual property law protects voice characteristics in certain commercial contexts. The reputational risk of being identified as having used an unconsented voice clone is real. **Buyer's responsibility is not a comfortable legal position.** If you are using voice cloning commercially in Australia, get legal advice before cloning any voice that isn't your own.

**Data goes offshore.** Audio and text data is processed on Fish Audio's US-hosted servers. No Australian region exists. For businesses handling private or sensitive content, this is a compliance question.

**Community library quality is uneven.** Two million voices sounds impressive. In practice, many community models are low-quality, inconsistent, or of uncertain consent origin. Finding a reliable, high-quality voice for a specific use case requires sorting through a lot of noise.

**Commercial rights require a paid plan.** The free tier is for personal use only. You need at least the Plus plan to use generated audio commercially — worth noting for freelancers or small businesses drawn in by the free tier.

**Support is thin.** Email support, documentation, and a Discord community. No AU-based support. No phone or live chat.

---

## Australian fit

Fish Audio's Australian fit is limited.

**Data:** US-hosted. No AU data residency. For businesses with privacy obligations — health, legal, finance — this is a barrier.

**Pricing:** USD only. GST applies to AU purchases. No AUD invoicing.

**Language and accents:** Fish Audio supports multiple English accents via community and custom voice models. There is no specific Australian English preset in the default library. Finding a convincing Australian accent requires either cloning your own voice or searching the community library — results vary.

**Legal risk:** The voice cloning consent issue has particular salience in Australia. With the Privacy Act 1988 covering biometric data and the broader legal framework around personality rights, the ease of cloning someone else's voice is not a feature — it is a liability. This is entirely on the user, not Fish Audio. Know the law before you proceed.

**Support:** No AU presence. Community-based support only.

---

## Alternatives

- [**ElevenLabs**](https://www.theaidirectory.com.au/elevenlabs-review-australia-2026/) — Higher price but better voice quality, a more curated voice library, and clearer consent controls on voice cloning. The professional choice if quality matters more than cost.
- [**Murf**](https://www.theaidirectory.com.au/murf-ai-review-australia-2026/) — A polished, interface-led TTS platform better suited to non-technical users. More expensive per minute but far easier to use for content creators.
- [**Resemble AI**](https://www.theaidirectory.com.au/resemble-ai-review-australia-2026/) — Another API-first TTS and voice cloning platform, with a stronger enterprise compliance story and clearer consent frameworks for cloned voices.

---

## Bottom line

Fish Audio is a technically capable, competitively priced TTS and voice cloning platform. For developers who need cheap, fast speech synthesis — particularly if they want to clone their own voice — it delivers.

The voice cloning consent situation is its defining limitation. The platform makes cloning any voice technically trivial. The legal and ethical responsibility sits entirely with the user. In Australia, that means understanding privacy law, IP law, and the practical risk of reputational damage if consent cannot be demonstrated.

For personal use or your own voice: worth trying. For commercial use in Australia involving any voice other than your own: get legal advice first.

---

## Score

| Criterion             | Score     | Notes                                                                                      |
| --------------------- | --------- | ------------------------------------------------------------------------------------------ |
| Value for Money       | 4/5       | Among the cheapest TTS APIs; good free tier; pay-as-you-go from the start                  |
| Ease of Use           | 3/5       | Web interface accessible; API well-documented; community library is noisy                  |
| Feature Depth         | 3/5       | S1 model is strong; voice cloning works; community library hit-and-miss                    |
| Australian Fit        | 1/5       | No AU data residency; USD only; consent risk elevated for AU users; no AU accents built-in |
| Reliability & Support | 2/5       | Fast API; no SLA documentation found; thin support; no AU presence                         |
| **Total**             | **13/25** | **⭐⭐⭐ Good**                                                                               |

*The ⭐⭐⭐⭐⭐ Excellent rating requires Australian Fit ≥ 4/5\. Fish Audio scores 1/5 on Australian Fit.* [*See our Methodology*](https://www.theaidirectory.com.au/methodology/)*.*

---

Update history 
- **1 Oct 2026 — re-verified by browser; negative confirmed.** `fish.audio/plan` on an Australian connection: Plus $5.50 and Pro $37.50 a month billed annually ($66 and $450 a year; $15 and $100 monthly), Free with 8,000 credits, and a new **Max plan at $749 a month billed annually** (25 million credits, 10 seats). The page's currency selector offers USD, JPY, EUR, KRW, GBP and BRL, but not AUD. The table's figures hold. `aud_native: false` already set. Score unchanged. Capture: `Dossiers/captures/fish-audio-pricing-2026-10-01.txt`. Wayback save queued, not yet cited.
- **21 Sep 2026** — **The grossed-up AUD column removed from the pricing table.** The table carried a second Australian column adding 10% GST to every converted figure, so each plan had two Australian prices and the higher one asserted a tax most of our readers do not pay: under the ATO rule our GST treatment rests on, a GST-registered Australian business that supplies its ABN is **not charged GST** on an imported digital service. **Nothing was re-cut** — the ex-GST column is unchanged and is the price. Removed, and kept visible here per our price-history rule: the column headed *"Approx AUD/month (if GST charged, +10%)"* (Plus **\~AUD $8.47**; Pro **\~AUD $57.75**). The GST position is stated in words once in the note above the pricing table. No USD price moved and no score changed. `reviewed:` untouched per our dating rule.

## Frequently asked questions

### Is voice cloning with Fish Audio legal in Australia?

Cloning your own voice with Fish Audio is legal. Cloning another person's voice without their explicit consent carries legal and reputational risk in Australia under the Privacy Act 1988 (which covers biometric data) and intellectual property law. Fish Audio's terms place full responsibility for obtaining consent on the user. Australian businesses using Fish Audio for voice cloning commercially should seek legal advice before cloning any voice other than their own.

### How much audio does Fish Audio need to clone a voice?

Fish Audio's S1 model can generate a voice clone from as little as 10 seconds of reference audio. Longer, cleaner recordings (30–60 seconds of studio-quality audio) produce better results. This is one of the lowest sample-length requirements in the market.

### How much does Fish Audio cost in Australian dollars?

Fish Audio prices in USD. The Plus plan is approximately USD $5.50/month (annual billing), roughly AUD $7.70/month ex-GST. API usage costs USD $15 per million UTF-8 bytes, approximately AUD $21.00/million bytes (ex-GST). GST (10%) applies to Australian purchases. A free tier with 8,000 monthly credits is available for personal use only. Confirm current pricing at fish.audio/plan.

---

## Sources

- Fish Audio Pricing & Plans: https://fish.audio/plan/ (accessed 26 June 2026) — [Wayback 18 August 2026](https://web.archive.org/web/20260818085149/https://fish.audio/plan/)
- Fish Audio API Pricing & Rate Limits: https://docs.fish.audio/developer-guide/models-pricing/pricing-and-rate-limits (accessed 26 June 2026) — [Wayback 2 July 2026](https://web.archive.org/web/20260702051237/https://docs.fish.audio/developer-guide/models-pricing/pricing-and-rate-limits)
- Fish Audio Voice Cloning: https://fish.audio/voice-clone/ (accessed 26 June 2026) — [Wayback 18 August 2026](https://web.archive.org/web/20260818085136/https://fish.audio/voice-clone/)
- Fish Audio S1 Model — TTS Arena Rankings: https://fish.audio/ (accessed 26 June 2026) — [Wayback 29 June 2026](https://web.archive.org/web/20260629234315/https://fish.audio/)
- Fish Speech GitHub (open-source model): https://github.com/fishaudio/fish-speech (accessed 26 June 2026) — [Wayback 4 July 2026](https://web.archive.org/web/20260704194535/https://github.com/fishaudio/fish-speech)
- Fish Audio Privacy Policy: https://fish.audio/privacy/ (accessed 26 June 2026) — [Wayback 29 August 2026](https://web.archive.org/web/20260829094823/https://fish.audio/privacy/)
- Fish Audio Blog — Voice Cloning Legal Guide 2026: https://fish.audio/blog/voice-cloning-guide/ (accessed 26 June 2026) — [Wayback 15 March 2026](https://web.archive.org/web/20260315174621/https://fish.audio/blog/voice-cloning-guide/)

---

*The AI Directory Australia is an independent editorial publication. We don't accept payment for positive reviews. Read our* [*Methodology*](https://www.theaidirectory.com.au/methodology/) *and* [*Affiliate Disclosure*](https://www.theaidirectory.com.au/affiliate-disclosure/)*.*

---