ElevenLabs Eleven v4 Turbo
Listed ElevenLabs
20
/100
ElevenLabs Eleven v4 Turbo is a text-to-speech (TTS) API from ElevenLabs, measured by one independent benchmark, Speko.
Speko
Google Chirp 3: HD Google · text-to-speech (TTS) API
Listed515.2 ms P50 time-to-first-audio on Coval; ranked 32nd of 112 in the TTS for Voice Agents 2026 index.
Benchmark summary: 18.5/100, ranked 32nd of 112 in the TTS for Voice Agents 2026 index (Listed). Strongest criterion: accuracy, 3.4 of 15 points.
Google Chirp 3: HD is a proprietary text-to-speech (TTS) API from Google, measured by two independent benchmarks: Artificial Analysis and Coval.
Overall score
18.5 /100
Score breakdown
- Real-time speed 10 /60
- Voice quality 4 /20
- Accuracy 3.4 /15
- Price 1.1 /5
Data checked on September 30, 2026
Updated after each benchmark capture
Every point comes from a public benchmark: sources · methodology
Google Chirp 3: HD ranks 32nd of 112 in the 2026 text-to-speech (TTS) API benchmark, with a score of 18.5/100.
Real-time latency
On Coval's 30-day window ending October 1, 2026 (11,855 runs), Google Chirp 3: HD has a P50 (median) time-to-first-audio (TTFA) of 515.2 ms, 27th lowest of the 31 models Coval measures. Its P99, the slowest 1% of turns, is 1,282.2 ms (27th lowest of the 31 models Coval measures). Latency consistency: a standard deviation of 225.8 ms (22nd lowest of the 31 models Coval measures). Mean time to first byte (TTFB) is 189.6 ms.
Accuracy (word error rate)
Coval measures an accuracy (word error rate) of 5% over 11,835 samples on the same window, 17th lowest of the 31 models Coval measures.
Voice quality
In the Artificial Analysis Speech Arena (capture of October 1, 2026), Google Chirp 3: HD has an Elo of 1,053 (±12, 95% confidence) from 3,296 blind comparisons, 56th highest of the 92 models Artificial Analysis rates.
Price
Artificial Analysis lists Google Chirp 3: HD at $30 per 1M characters (43rd lowest of the 78 models Artificial Analysis lists a price for).
Languages tested
Tested in Japanese, Hindi, Tamil, Telugu and Thai by Artificial Analysis or Speko. A language missing here was not tested, which does not mean it is unsupported.
Benchmark measurements
Latency and accuracy — Coval (30-day window)
- Mean TTFA
- 542.5 ms
- P50 / median time-to-first-audio (TTFA)
- 515.2 ms
- P90 TTFA
- 814.3 ms
- P95 TTFA
- 899.2 ms
- P99 TTFA (slowest turns)
- 1,282.2 ms
- Latency consistency (standard deviation)
- 225.8 ms
- Leading silence
- 352.9 ms
- Time to first byte (network arrival)
- 189.6 ms
- Accuracy: word error rate (WER)
- 5%
- WER basis
- per-clip mean
- Latency runs
- 11,855 runs
- Accuracy runs
- 11,835 runs
- Window
- 30d
- Coval snapshot
- 2026-10-01T10:11:47.318752Z
- Coval model ID
- chirp-3-hd
- Coval model page
- benchmarks.coval.ai
Artificial Analysis Speech Arena
- Speech Arena Elo (blind listener preference)
- 1,053
- Elo 95% interval (±)
- 12
- Arena rank
- 57
- Arena rank range
- 49-62
- Arena appearances
- 3,296
- Voices tested
- 8
- Price per 1M characters
- 30 USD
- Released
- Mar 2025
- Artificial Analysis model
- Chirp 3: HD
- Artificial Analysis leaderboard
- artificialanalysis.ai
Not collected
- Speed (Speko time to first audio) — not measured by Speko (capture of 2026-10-01); counts 0
- Quality (Speko naturalness) — not measured by Speko (capture of 2026-10-01); counts 0
- Accuracy (Speko robustness) — not measured by Speko (capture of 2026-10-01); counts 0
- Price (Speko cost per 1M characters) — not measured by Speko (capture of 2026-10-01); counts 0
Similar TTS models
Nari Qwen3-TTS
Listed Open weights · Alibaba (Qwen)
14.6
/100
Nari Qwen3-TTS is an open-weights text-to-speech (TTS) API from Alibaba (Qwen), served by Nari Labs, measured by one independent benchmark, Speko.
Speko
SpeechifyAI Simba 3.0
Listed Proprietary · Speechify
26
/100
SpeechifyAI Simba 3.0 is a proprietary text-to-speech (TTS) API from Speechify, measured by two independent benchmarks: Artificial Analysis and Coval.
Spanish German French Portuguese
+5