Speech pricing (VoiceThere STT & TTS)
When your project uses VoiceThere speech for speech-to-text or text-to-speech, VoiceThere bills usage credits from the runner usage meter at session end. Cloud speech vendors (OpenAI, ElevenLabs, Fish Audio, and others) are billed by your vendor — these rates do not apply.
Cash equivalents use the catalog rate of €0.001 and $0.00125 per credit. See also Billing & usage credits and the public pricing page.
How STT is measured
STT quantity is audio seconds of user speech sent to the recognizer, measured from user_speaking_start through user_speaking_end (each utterance is rounded up to whole seconds before summing). The rate is 5 credits per minute (about €0.005/min or $0.0063/min at catalog cash per credit).
How TTS is measured
TTS quantity is the Unicode character count of text passed to the runner speak(text) API (spaces and punctuation count). Credits depend on the voice quality tier selected in the VoiceThere TTS catalog — not playback duration.
Replies replayed from the phrase cache are not billed: only text that is actually synthesised counts toward TTS characters.
Quality tiers
| Quality | Credits / 1,000 chars | ≈ € / 1,000 | ≈ $ / 1,000 |
|---|---|---|---|
| low | 5 | €0.0050 | $0.0063 |
| medium | 15 | €0.0150 | $0.0187 |
| high | 100 | €0.1000 | $0.1250 |
Quality is inferred from the Piper bundle name (-low, -medium, -high) or defaults to medium for unlabeled voices.
Choosing a voice quality
Every VoiceThere voice belongs to one of three quality tiers. The tier sets the price per 1,000 characters and how quickly the agent starts speaking.
| Quality | Credits / 1,000 chars | Time to first audio | Good for |
|---|---|---|---|
| Low | 5 | about 0.1–0.3 s | The lowest cost and quick replies, for high call volumes and short prompts. |
| Medium | 15 | about 0.1–0.3 s | The recommended default: clear, natural speech at a moderate price. |
| High | 100 | about 0.4–1.0 s | Use cases where the sound of the voice matters more than speed or cost. |
Start with a medium voice; the default English voice (Amy, medium) is one. High-quality voices need about eight times the compute of a medium voice for the same text. That is why they cost more, and why the first word of each reply arrives later. Callers can hear that pause, so choose a high voice only when its sound matters to your use case.
Time to first audio is the time from the end of the caller's turn to the first audio of the reply, measured on VoiceThere servers with ten conversations at once. Network time to the caller comes on top.
VoiceThere TTS voice catalog
Each catalog voice maps to one quality tier. Changing the project TTS model changes both synthesis and the TTS credit rate.
| Voice | Catalog id | Quality | Cr / 1k | ≈ € / 1k | ≈ $ / 1k |
|---|---|---|---|---|---|
| English (Amy, Piper medium) | en | medium | 15 | €0.0150 | $0.0187 |
| English (Amy, Piper low) | en-amy-low | low | 5 | €0.0050 | $0.0063 |
| English (Lessac, Piper medium) | en-lessac | medium | 15 | €0.0150 | $0.0187 |
| English (Lessac, Piper high) | en-lessac-high | high | 100 | €0.1000 | $0.1250 |
| English (Lessac, Piper low) | en-lessac-low | low | 5 | €0.0050 | $0.0063 |
| English (GLaDOS) | en-glados | medium | 15 | €0.0150 | $0.0187 |
| English (GLaDOS high) | en-glados-high | high | 100 | €0.1000 | $0.1250 |
| English (Ryan, Piper medium) | en-ryan | medium | 15 | €0.0150 | $0.0187 |
| English (Ryan, Piper high) | en-ryan-high | high | 100 | €0.1000 | $0.1250 |
| English (Ryan, Piper low) | en-ryan-low | low | 5 | €0.0050 | $0.0063 |
| English (Joe, Piper medium) | en-joe | medium | 15 | €0.0150 | $0.0187 |
| English (Bryce, Piper medium) | en-bryce | medium | 15 | €0.0150 | $0.0187 |
| English (John, Piper medium) | en-john | medium | 15 | €0.0150 | $0.0187 |
| English (Kristin, Piper medium) | en-kristin | medium | 15 | €0.0150 | $0.0187 |
| English (Norman, Piper medium) | en-norman | medium | 15 | €0.0150 | $0.0187 |
| English (Danny, Piper low) | en-danny-low | low | 5 | €0.0050 | $0.0063 |
| English (Kathleen, Piper low) | en-kathleen-low | low | 5 | €0.0050 | $0.0063 |
| English (LJSpeech, Piper medium) | en-ljspeech | medium | 15 | €0.0150 | $0.0187 |
| English (LJSpeech, Piper high) | en-ljspeech-high | high | 100 | €0.1000 | $0.1250 |
| English (Arctic multi-speaker, Piper medium) | en-arctic | medium | 15 | €0.0150 | $0.0187 |
| English (HFC female, Piper medium) | en-hfc-female | medium | 15 | €0.0150 | $0.0187 |
| English (HFC male, Piper medium) | en-hfc-male | medium | 15 | €0.0150 | $0.0187 |
| English GB (Alan, Piper medium) | en-gb-alan | medium | 15 | €0.0150 | $0.0187 |
| English GB (Cori, Piper medium) | en-gb-cori | medium | 15 | €0.0150 | $0.0187 |
| English GB (Jenny, Piper medium) | en-gb-jenny | medium | 15 | €0.0150 | $0.0187 |
| Spanish (GLaDOS medium) | es | medium | 15 | €0.0150 | $0.0187 |
| German (Thorsten medium) | de | medium | 15 | €0.0150 | $0.0187 |
| German (Thorsten high) | de-thorsten-high | high | 100 | €0.1000 | $0.1250 |
| German (GLaDOS medium) | de-glados | medium | 15 | €0.0150 | $0.0187 |
| French (Siwis, Piper medium) | fr | medium | 15 | €0.0150 | $0.0187 |
| Italian (Paola, Piper medium) | it | medium | 15 | €0.0150 | $0.0187 |
| Portuguese BR (Faber, Piper medium) | pt | medium | 15 | €0.0150 | $0.0187 |
| Dutch (Pim, Piper medium) | nl | medium | 15 | €0.0150 | $0.0187 |
| Polish (Gosia, Piper medium) | pl | medium | 15 | €0.0150 | $0.0187 |
| Russian (Irina, Piper medium) | ru | medium | 15 | €0.0150 | $0.0187 |
| Hindi (Priyamvada, Piper medium) | hi | medium | 15 | €0.0150 | $0.0187 |
| Chinese + English (Melo TTS) | zh | high | 100 | €0.1000 | $0.1250 |