click any column header to sort
← SpeedSpeedSamples → · all runs

TTS Bench — Samples — linux-default

Rig: linux-3090 — AMD Ryzen 9 5900XT 16-Core Processor · NVIDIA GeForce RTX 3090 24GB · 63 GB RAM · Linux 6.8.0-117-generic
Label: default voice
5 prompt(s) · one section per prompt · all models ranked by warm TTFA (fastest first) within each
Each prompt section shows every model's audio output, ordered by warm TTFA (fastest first). Click any audio player to hear that model's rendering.

Prompt 1

[en]"Open the browser and read my email."
Rank Model Device TTFA warm Audio
1 Kokoro cuda 47ms
2 Piper cpu 62ms
3 Pocket-TTS cpu 85ms
4 LuxTTS cuda 194ms
5 NeuTTS Nano cuda 315ms
6 Kokoro cpu 361ms
7 NeuTTS Nano cpu 392ms
8 Soprano 80M cuda 397ms
9 NeuTTS Air cuda 486ms
10 Chatterbox Turbo cuda 521ms
11 NeuTTS Air cpu 546ms
12 Coqui XTTS-v2 cuda 557ms
13 Supertonic cpu 596ms
14 Soprano 80M cpu 662ms
15 Qwen3-TTS 1.7B (CUDA-graph) cuda 679ms
16 OmniVoice cuda 713ms
17 KittenTTS cpu 736ms
18 Chatterbox cuda 1.06s
19 VibeVoice Realtime 0.5B cuda 1.08s
20 LuxTTS cpu 1.10s
21 F5-TTS cuda 1.11s
22 Magpie-TTS cuda 1.42s
23 VibeVoice 1.5B cuda 1.46s
24 MOSS-TTS-Nano cuda 1.66s
25 Qwen3-TTS 1.7B cuda 2.08s
26 VoxCPM2 2B cuda 2.36s
27 IndexTTS-2 cpu 2.65s
28 IndexTTS-2 cuda 2.88s
29 Coqui XTTS-v2 cpu 3.07s
30 Sesame CSM-1B cuda 3.24s
31 MOSS-TTS-Nano cpu 3.38s
32 Chatterbox Turbo cpu 3.79s
33 Dia 1.6B cuda 4.12s
34 VibeVoice Realtime 0.5B cpu 5.46s
35 Chatterbox cpu 6.02s
36 ZipVoice 123M cpu 7.14s
37 OmniVoice cpu 7.73s
38 Magpie-TTS cpu 8.50s
39 Qwen3-TTS 1.7B cpu 13.70s
40 VoxCPM2 2B cpu 13.71s
41 Sesame CSM-1B cpu 14.84s
42 VibeVoice 1.5B cpu 19.39s
43 Mars5-TTS cpu 37.79s
44 Mars5-TTS cuda 37.89s
45 F5-TTS cpu 38.91s

Prompt 2

[en]"I'll start a new git branch, push the changes, and open a pull request when the tests pass."
Rank Model Device TTFA warm Audio
1 Kokoro cuda 63ms
2 Pocket-TTS cpu 110ms
3 Piper cpu 150ms
4 LuxTTS cuda 205ms
5 NeuTTS Nano cuda 315ms
6 NeuTTS Nano cpu 390ms
7 NeuTTS Air cuda 462ms
8 NeuTTS Air cpu 547ms
9 OmniVoice cuda 678ms
10 Kokoro cpu 824ms
11 Supertonic cpu 1.00s
12 Soprano 80M cuda 1.01s
13 Chatterbox Turbo cuda 1.12s
14 Coqui XTTS-v2 cuda 1.17s
15 F5-TTS cuda 1.34s
16 LuxTTS cpu 1.57s
17 Qwen3-TTS 1.7B (CUDA-graph) cuda 1.59s
18 Soprano 80M cpu 1.75s
19 Chatterbox cuda 1.90s
20 KittenTTS cpu 1.91s
21 VibeVoice Realtime 0.5B cuda 2.24s
22 VoxCPM2 2B cuda 2.43s
23 VibeVoice 1.5B cuda 2.51s
24 MOSS-TTS-Nano cuda 2.86s
25 Magpie-TTS cuda 3.12s
26 IndexTTS-2 cpu 3.84s
27 IndexTTS-2 cuda 4.27s
28 MOSS-TTS-Nano cpu 5.18s
29 Qwen3-TTS 1.7B cuda 5.54s
30 Sesame CSM-1B cuda 6.62s
31 Chatterbox Turbo cpu 6.65s
32 Coqui XTTS-v2 cpu 7.15s
33 VibeVoice Realtime 0.5B cpu 11.00s
34 Chatterbox cpu 12.08s
35 OmniVoice cpu 12.82s
36 ZipVoice 123M cpu 13.71s
37 Dia 1.6B cuda 21.78s
38 VoxCPM2 2B cpu 21.93s
39 Magpie-TTS cpu 23.70s
40 Qwen3-TTS 1.7B cpu 26.55s
41 VibeVoice 1.5B cpu 31.43s
42 Mars5-TTS cuda 47.97s
43 Mars5-TTS cpu 47.98s
44 F5-TTS cpu 49.20s
45 Sesame CSM-1B cpu 61.47s

Prompt 3

[en]"The Parakeet TDT zero point six billion parameter model achieves one point six nine percent word error rate on LibriSpeech test-clean, beating Whisper Large V3 at two point seven percent while running at over two thousand times realtime on a single GPU."
Rank Model Device TTFA warm Audio
1 Kokoro cuda 133ms
2 Pocket-TTS cpu 138ms
3 LuxTTS cuda 257ms
4 NeuTTS Nano cuda 348ms
5 NeuTTS Nano cpu 419ms
6 Piper cpu 428ms
7 NeuTTS Air cuda 492ms
8 NeuTTS Air cpu 568ms
9 OmniVoice cuda 1.05s
10 F5-TTS cuda 1.94s
11 Supertonic cpu 2.67s
12 Soprano 80M cuda 2.99s
13 Chatterbox Turbo cuda 3.19s
14 Kokoro cpu 3.30s
15 LuxTTS cpu 3.33s
16 KittenTTS cpu 3.90s
17 Coqui XTTS-v2 cuda 4.15s
18 Qwen3-TTS 1.7B (CUDA-graph) cuda 4.32s
19 Chatterbox cuda 4.60s
20 Soprano 80M cpu 4.63s
21 VibeVoice Realtime 0.5B cuda 6.56s
22 VoxCPM2 2B cuda 7.01s
23 MOSS-TTS-Nano cuda 7.14s
24 Magpie-TTS cuda 9.82s
25 VibeVoice 1.5B cuda 9.97s
26 IndexTTS-2 cpu 10.97s
27 IndexTTS-2 cuda 11.19s
28 MOSS-TTS-Nano cpu 12.23s
29 Sesame CSM-1B cuda 12.95s
30 Qwen3-TTS 1.7B cuda 14.99s
31 Dia 1.6B cuda 18.82s
32 Chatterbox Turbo cpu 22.73s
33 Coqui XTTS-v2 cpu 25.88s
34 OmniVoice cpu 31.64s
35 Chatterbox cpu 31.83s
36 VibeVoice Realtime 0.5B cpu 31.98s
37 VoxCPM2 2B cpu 59.59s
38 Qwen3-TTS 1.7B cpu 72.20s
39 F5-TTS cpu 83.01s
40 Sesame CSM-1B cpu 84.06s
41 VibeVoice 1.5B cpu 95.89s
42 Mars5-TTS cpu 96.12s
43 Mars5-TTS cuda 97.23s
44 Magpie-TTS cpu 106.38s

Prompt 4

[en]"Run pytest tests slash test underscore voice dot py with verbose flag and capture flag set to no."
Rank Model Device TTFA warm Audio
1 Kokoro cuda 70ms
2 Pocket-TTS cpu 123ms
3 Piper cpu 152ms
4 LuxTTS cuda 206ms
5 NeuTTS Nano cuda 323ms
6 NeuTTS Nano cpu 406ms
7 NeuTTS Air cuda 473ms
8 NeuTTS Air cpu 555ms
9 OmniVoice cuda 672ms
10 Kokoro cpu 959ms
11 Supertonic cpu 1.12s
12 Soprano 80M cuda 1.15s
13 Chatterbox Turbo cuda 1.34s
14 F5-TTS cuda 1.34s
15 LuxTTS cpu 1.62s
16 KittenTTS cpu 1.71s
17 Soprano 80M cpu 1.88s
18 Coqui XTTS-v2 cuda 1.99s
19 Qwen3-TTS 1.7B (CUDA-graph) cuda 2.07s
20 Chatterbox cuda 2.26s
21 MOSS-TTS-Nano cuda 3.08s
22 VibeVoice Realtime 0.5B cuda 3.22s
23 VoxCPM2 2B cuda 3.59s
24 Magpie-TTS cuda 4.32s
25 VibeVoice 1.5B cuda 4.81s
26 IndexTTS-2 cpu 5.23s
27 IndexTTS-2 cuda 5.47s
28 MOSS-TTS-Nano cpu 5.95s
29 Qwen3-TTS 1.7B cuda 7.63s
30 Chatterbox Turbo cpu 9.17s
31 Dia 1.6B cuda 10.60s
32 Sesame CSM-1B cuda 11.84s
33 OmniVoice cpu 12.84s
34 Coqui XTTS-v2 cpu 13.26s
35 VibeVoice Realtime 0.5B cpu 14.99s
36 ZipVoice 123M cpu 15.03s
37 Chatterbox cpu 16.17s
38 Magpie-TTS cpu 33.05s
39 VoxCPM2 2B cpu 33.64s
40 Qwen3-TTS 1.7B cpu 36.51s
41 Mars5-TTS cpu 50.17s
42 F5-TTS cpu 50.71s
43 Mars5-TTS cuda 51.80s
44 VibeVoice 1.5B cpu 62.28s
45 Sesame CSM-1B cpu 70.00s

Prompt 5

[fr]"Bonjour, je m'appelle Cicero et je vais vous aider avec votre code aujourd'hui."
Rank Model Device TTFA warm Audio
1 Kokoro cuda 49ms
2 Piper cpu 109ms
3 NeuTTS Nano cuda 257ms
4 Pocket-TTS cpu 283ms
5 NeuTTS Nano cpu 338ms
6 Kokoro cpu 668ms
7 OmniVoice cuda 683ms
8 Coqui XTTS-v2 cuda 923ms
9 Supertonic cpu 993ms
10 Qwen3-TTS 1.7B (CUDA-graph) cuda 1.41s
11 VoxCPM2 2B cuda 1.82s
12 MOSS-TTS-Nano cuda 3.06s
13 Magpie-TTS cuda 3.32s
14 MOSS-TTS-Nano cpu 3.71s
15 Qwen3-TTS 1.7B cuda 4.80s
16 Coqui XTTS-v2 cpu 6.18s
17 OmniVoice cpu 11.73s
18 ZipVoice 123M cpu 12.88s
19 VoxCPM2 2B cpu 16.92s
20 Magpie-TTS cpu 27.26s
21 Qwen3-TTS 1.7B cpu 27.66s