Snapshot
Continue to About ↓- What it does
- Gradium is a real-time voice AI API that developers use to add speech to applications: text-to-speech, streaming speech-to-text with semantic turn detection, voice-to-voice and speech-to-text translation, instant and pro voice cloning, and voice design from a text prompt.
More detail
- It also publishes Phonon, an open on-device TTS model.
- Best for
- Developers and product teams building real-time voice agents, phone agents and customer-care bots that need streaming speech-to-text and low-latency text-to-speech through one API.
More detail
- Teams handling noisy or telephone-quality audio, and enterprises that need SOC 2 Type II, ISO 27001, EU or US data residency and zero data retention.
- Pricing
- Credit-based subscription with an optional pay-as-you-go overage.
More detail
- Plans: Free at $0 with 45k credits per month (text-to-speech, speech-to-text, voice design, no commercial use)
- XS at $12/mo with 225k credits (commercial use, 1,000 voices, EU or US residency)
- S at $39/mo with 900k credits (cheaper
- Main limitation
- Teams that need broad language coverage, since Gradium currently lists only English, French, Spanish, Portuguese and German.
More detail
- Buyers who need long uninterrupted single sessions should check the session duration cap with Gradium first.
- Record
- Living and monitored
- Reviewed
- Evidence
- Public web research
- Machine access
- Structured profile ↗
About
- Gradium is a real-time voice AI API that developers use to add speech to applications: text-to-speech, streaming speech-to-text with semantic turn detection, voice-to-voice and speech-to-text translation, instant and pro voice cloning, and voice design from a text prompt.
- It also publishes Phonon, an open on-device TTS model.
What it helps with
- Adding natural-sounding speech output to voice agents without building and hosting a TTS model.
- Handling conversational turn-taking with semantic detection of when a user has finished speaking, instead of fixed silen
- Transcribing telephone-quality and noisy audio in streaming form.
- Improving recognition of domain words through custom vocabulary keyword boosting without retraining a model.
- Creating a usable voice clone quickly from a short audio sample, or a higher-fidelity pro clone from longer audio.
- Creating a new voice with no audio sample, from a written description.
- Translating speech voice-to-voice in real time.
- Meeting enterprise security and data-handling requirements through SOC 2 Type II, ISO 27001, a DPA and zero data retenti
Capabilities
A practical view of what is available and where limitations apply.
| Offering | Available | Detail |
|---|---|---|
| Text-to-Speech | Yes | Turn text into speech; billed at 1 credit per character. |
| Speech-to-Text | Yes | Streaming transcription over WebSocket with semantic turn detection, keyword boosting and tunable latency; billed at 3 credits per second. |
| Live Translation | Yes | Speech-to-speech translation at 30 credits per second and speech-to-text translation at 4 credits per second. |
| Voice Cloning | Yes | Instant clones from roughly 10 seconds of audio; Pro clones require at least 30 minutes of clean audio and are included from the M plan. Explicit consent of the voice owner is required and is the customer's responsibility. |
| Voice Design | Yes | Generate up to five candidate voices from a description of up to 500 characters, then save the chosen one as a permanent voice ID. |
| Phonon on-device TTS model | Yes | Described by Gradium as its first open, on-device TTS model. |
| Language coverage | Limited | English, French, Spanish, Portuguese and German; more languages are stated to be in development. |
| Speaker diarization | Limited | Stated to be in beta and to work for two speakers today. |
| Confidence scores | No | Stated to be in development. |
| Session length | Limited | The pricing FAQ states a maximum session of 300 seconds and advises splitting longer text or audio; a separate docs FAQ states 3000 seconds, so the limit should be confirmed with Gradium. |
Bot trust
How confidently AI systems can verify this profile from public evidence.
—of 100
ChatGPT—/100
Claude—/100
This profile has not completed a Bot Score assessment yet.
Evidence
Where this record came from, when it was reviewed, and how confidently it can be used.
- Verification
- Scribble researched these facts from the linked public sources.
- The owner has not manually confirmed every field.
- Last reviewed
- Primary source
- gradium.ai
Technical profile details
Record ID
3863cd94-aa81-44bb-b760-8d6554bf6168Canonical URLhttps://profiles.scribble.network/listing/gradium-3863cd94