Fish Audio vs Resemble AI: pricing, features and which to choose

Fish Audio began as Fish Speech, a project by former Nvidia researcher Shijia Liao, who trained a voice-generation model on one GPU and open-sourced it. The company later rebranded as Fish Audio, with Rissa Cao serving as CEO and co-founder. Its public history records the first Fish Speech release in December 2023. The company launched five models in the year before July 2026, including four speech-generation systems and one speech-to-text model. TechCrunch reported that Fish Audio raised a $52 million seed round in July 2026, led by Coreline Ventures and Capital Today. The report also cited more than eight million users across hosted and open-source models, $21 million in annual recurring revenue, and a GitHub repository exceeding 31,000 stars.

A typical Fish Audio workflow begins by selecting a voice or creating one through cloning or Voice Design, then entering a script and adding bracketed directions such as [whisper] or [laugh]. Users generate audio, review it, and download or share the result. For production integrations, developers send text to the REST TTS endpoint or use WebSocket streaming through Python or TypeScript SDKs. S2.1-Pro supports multi-speaker dialogue and 83 languages; s2.1-pro-free suits development under fair-use limits. For transcription, teams upload a recording to /v1/asr and specify the transcribe-1-pro header to receive speaker markers, turns, timestamps, and vocal-event cues. Live applications instead keep a realtime WebSocket open with transcribe-stream-1, billed by session duration.

Fish Audio differs from ElevenLabs, Cartesia, and PlayHT in its combination of expressive controls, open models, and deployment options. S2.1 Pro offers 83 languages, multi-speaker dialogue, and free development access, while bracket-based direction and open-weight models provide control over delivery and deployment. The $15-per-million-UTF-8-byte API price differs from credit-based web plans, complicating budgeting across products. Starter API accounts allow five concurrent requests, and long recordings can occupy a slot for minutes. Its terms permit use of Usage Data and Content to develop or enhance models, while voice submissions and public content carry licensing and consent responsibilities. Fish Audio suits creators and engineering teams prioritizing expressive control, multilingual output, and self-hosting.

Resemble AI is a Toronto and San Francisco company founded in 2019 by Zohaib Ahmed and Saqib Muhammad to make synthetic voices for game studios. For five years it sold voice cloning, text-to-speech and speech-to-speech through a web studio and API, raising an $8 million round in July 2023 and later a $13 million round aimed at AI-generated threats. In August 2026 Ahmed published a post titled "Why a former voice AI company went all in on deepfake detection" confirming the company had stopped selling voice generation to new customers. Existing voice customers are still supported. Resemble also retired more than 300 older blog posts on voice cloning, which is why many old links now land on that announcement. Anyone researching Resemble as a voice generator should treat it as a detection vendor with a voice research arm.

The current product is a set of five APIs. Detect returns a verdict on whether audio, video or an image is synthetic in under 300 milliseconds, according to the company. Watermarker embeds an inaudible or invisible mark at the point of generation and reads it back after compression and re-encoding, with C2PA support. Identity enrolls a known voice and face, then checks whether an incoming call or video matches it. Signal matches calls against libraries of known fraud patterns without storing transcripts, and Intelligence turns a detection score into plain-language reasoning and JSON. Ready-made products sit on top: meeting protection for Microsoft Teams, a Chrome extension that flags deepfakes in the browser, and agent detection for spotting AI callers. A typical deployment pipes contact-center audio or uploaded media through Detect and routes flagged items to an analyst.

The main limitation for most readers is that Resemble no longer sells the product it was known for. Teams that want hosted voice cloning should look at ElevenLabs, which offers instant and professional cloning from $5 a month, or Respeecher for studio-grade speech-to-speech. Developers comfortable running models themselves can use Chatterbox from Resemble's own repository at no cost, but without a commercial voice service behind it. On detection, Resemble competes with Reality Defender and Pindrop, which also sell real-time deepfake screening to banks and call centers; Resemble's edge is that it publishes self-serve prices and an API key you can get without a sales call. Pick Resemble if you need to screen calls, meetings or uploads for synthetic media. Look elsewhere if you need to generate voices.