Try AnyToSpeech
Benchmark › Methodology
Speech test set speech-v2 · music test set music-v2 · last run 2026-11-22

Methodology

Everything needed to check or repeat these results: the models and settings, every test item, every scoring rule, and the places where the method is weak.

Principles

Speech models tested

ModelAPI modelVoiceReached throughEmotion control used
Qwen3 TTS
Alibaba
qwen3-ttsSerenaReplicateText instruction
AnyToSpeech Standard voices
AnyToSpeech
site pipeline on gemini-3.8-flash-ttsKoreAnyToSpeech pipelineText instruction
AnyToSpeech Premium and Characters
AnyToSpeech
site pipeline on inworld-tts-2AshleyAnyToSpeech pipelineText instruction
Eleven v3
ElevenLabs
eleven_v3SarahReplicateInline tag
Eleven Flash v2.5
ElevenLabs
eleven_flash_v2_5SarahReplicateNone offered
Eleven Multilingual v2
ElevenLabs
eleven_multilingual_v2SarahReplicateNone offered
Gemini 3.8 Flash-Lite TTS
Google
gemini-3.8-flash-lite-ttsKoreGemini APIText instruction
Gemini 3.8 Flash TTS
Google
gemini-3.8-flash-ttsKoreGemini APIText instruction
Gemini 2.5 Pro TTS
Google
gemini-2.5-pro-preview-ttsKoreGemini APIText instruction
Kokoro 82M
Hexgrad (open source)
kokoro-82maf_bellaReplicateNone offered
Inworld TTS-2
Inworld
inworld-tts-2AshleyInworld APIText instruction
Inworld TTS-1.5 Max
Inworld
inworld-tts-1.5-maxAshleyInworld APINone offered
MiniMax Speech 2.8 HD
MiniMax
speech-2.8-hdEnglish_WiseladyReplicateEmotion setting
MiniMax Speech 2.8 Turbo
MiniMax
speech-2.8-turboEnglish_WiseladyReplicateEmotion setting
GPT-4o mini TTS
OpenAI
gpt-4o-mini-ttsalloyOpenAI APIText instruction
TTS-1
OpenAI
tts-1alloyOpenAI APINone offered
TTS-1 HD
OpenAI
tts-1-hdalloyOpenAI APINone offered
Chatterbox Multilingual
Resemble AI (open source)
chatterbox-multilingualdefaultReplicateNone offered
Grok TTS
xAI
grok-text-to-speecheveReplicateInline tag

Transcripts come from two recognisers (whisper-1 and gpt-4o-transcribe); the transcript with fewer errors is used for each clip. Models reached through Replicate run on Replicate's hosted version of the vendor's model.

Speech tests and scoring

Word accuracy 25% of overall

Test: 60 everyday English sentences with no numbers or unusual names.
Scoring: Each clip is transcribed by two speech recognisers and compared word for word with the text. We count missing words, added words and wrong words, and keep the transcript with fewer errors so a recogniser mistake is not blamed on the voice. Score = 100 minus the word error rate.

Numbers, dates and symbols 20% of overall

Test: 60 sentences, each with one trap: $1,250.50, 03/14/2025, 7:45 PM, 12.5%, -5°C, Dr. and St., an email address, a URL, a flight number, a version number and so on.
Scoring: A sentence passes when the transcript shows the item was spoken correctly, using a published list of accepted readings for each item. Score = share of the 60 that pass. Where a model has a documented text-normalisation switch, it is turned on.

Long text 15% of overall

Test: One 15,000-character document, sent at five lengths: about 1,000, 2,000, 4,000, 8,000 and 15,000 characters. Each length is a single request with no splitting on our side, so every model is pushed up to and past its own limit.
Scoring: Each recording is transcribed in one-minute pieces and compared word for word with the text sent. A length scores its word accuracy; a length the model rejects, cuts short or garbles scores what comes back, which is 0 when nothing usable does. Score = average across the five lengths. "Longest clean read" is the largest length read with at least 97% of words correct.

Repeated words 10% of overall

Test: 20 lines built on repetition: "No, no, no, no, no. I said no.", a line that repeats a whole sentence, a spoken code word given twice.
Scoring: Word-by-word comparison. Speech recognisers themselves drop repeated sentences, so each clip is also cut at its sentence pauses and every piece is transcribed separately; the best transcript counts.

Other languages 15% of overall

Test: 8 sentences each in Spanish, French, German, Italian, Portuguese, Hindi, Japanese and Mandarin, read with the model's default voice (or its voice for that language when the model requires one).
Scoring: Word error rate per language (character error rate for Japanese and Mandarin). Score = average accuracy across the languages the model supports; the number of supported languages is shown next to it.

Emotion control 15% of overall

Test: 10 neutral sentences, each requested angry, excited, scared and whispered, through the model's own documented control (instruction field, inline tag or emotion setting). Models with no such control are rendered plain and scored the same way.
Scoring: Measured against the same model's neutral reading of the same sentence. Angry, excited and scared pass when median pitch rises by at least 2 semitones; a whisper passes when voiced sound drops by at least half. This checks that the delivery changed in the right direction. It does not prove a listener would name that exact emotion; blind listener voting is planned for that.

Reliability not in overall

Test: Every request in the benchmark except the long-text ladder, about 255 per model.
Scoring: A request is clean when the first attempt returns audio that is not silent. Rate-limit responses caused by the test itself are retried and not counted. Failures are retried once; an item that fails twice counts as fully wrong in its own category.

Speed not in overall

Test: The 60 word-accuracy sentences.
Scoring: Median generation time, and audio length divided by generation time ("times real time"). For models called directly this is the full request time. For models reached through Replicate it is the compute time Replicate reports, which leaves out queueing. The two are not directly comparable, so speed is shown but never counted in the overall score.

A request that fails twice counts as fully wrong in its category. Differences under about 2 points between models are within the noise of the recognisers and should be read as a tie.

Speech test items

Word accuracy (60 sentences)

  1. The ferry leaves the harbor every morning just after sunrise, and most of the passengers are still half asleep.
  2. She folded the letter twice, slipped it into her coat pocket, and walked out without saying a word.
  3. If the weather holds, we should reach the summit before noon and be back at camp by dark.
  4. Nobody in the village could remember a winter as long or as cold as that one.
  5. The engineer checked every bolt on the bridge before she allowed the first truck to cross.
  6. He opened the old book carefully, because the pages were thin and the binding had started to crack.
  7. Our neighbors planted tomatoes, peppers, and basil along the fence behind their garage.
  8. When the lights went out, the children laughed and reached for the candles on the shelf.
  9. The committee agreed to postpone the vote until every member had read the full report.
  10. A small brown dog followed us all the way from the bakery to the edge of the park.
  11. You can leave your shoes by the door and hang your jacket on the hook beside the mirror.
  12. The museum stays open late on Fridays, which is the only time I can visit after work.
  13. Although the recipe looked simple, the bread came out of the oven flat and far too salty.
  14. They spent the whole afternoon repairing the roof before the next storm rolled in from the coast.
  15. Please remind your brother that the library books are due back before the weekend.
  16. The pilot announced a short delay, thanked everyone for their patience, and switched off the microphone.
  17. I would rather take the slow train through the mountains than fly and miss the view.
  18. Her grandmother taught her to knit when she was seven, and she has never stopped since.
  19. The river was too wide to swim across, so we walked upstream until we found a bridge.
  20. After the concert, the musicians packed their instruments and drove through the night to the next city.
  21. The surgeon explained the procedure twice and then asked whether the family had any questions.
  22. A thin layer of frost covered the windshield, so he scraped it clear with an old plastic card.
  23. Every spring the orchard fills with bees, and the farmer walks the rows to count the blossoms.
  24. She had rehearsed the speech for weeks, yet her hands still shook when she reached the stage.
  25. The detective noticed that the kitchen clock had stopped at exactly the moment the power failed.
  26. We rented a small cabin by the lake and spent the evenings reading beside the fire.
  27. The architect proposed a taller building, but the council rejected the plan after a long debate.
  28. His youngest daughter wants to study marine biology and spend her life near the ocean.
  29. The bakery on the corner sells out of fresh bread long before most people wake up.
  30. Before leaving the laboratory, the students washed the glassware and labeled each sample.
  31. The orchestra tuned for a moment, the conductor raised his arms, and the hall fell silent.
  32. I forgot my umbrella at the restaurant, and by the time I returned it had vanished.
  33. The mechanic wiped his hands, shook his head, and told us the engine could not be saved.
  34. Thousands of birds gather in the marsh each autumn before they begin the journey south.
  35. She translated the entire novel herself because no publisher was willing to pay for it.
  36. The lawyer argued that the contract was invalid because nobody had signed the final page.
  37. On quiet nights you can hear the freight trains crossing the valley from our porch.
  38. The teacher divided the class into groups and gave each one a different chapter to present.
  39. He paid for the groceries, carried the bags to the car, and realized he had lost his keys.
  40. Scientists believe the volcano has been dormant for centuries, though small tremors continue.
  41. The gardener trimmed the hedge so evenly that it looked like a green stone wall.
  42. My cousin moved abroad last year and now sends long letters instead of calling.
  43. The judge listened to both sides, reviewed the evidence, and delivered a brief verdict.
  44. A sudden gust of wind lifted the tent from the ground and carried it across the field.
  45. The nurse dimmed the lights, checked the monitor once more, and quietly closed the door.
  46. They built the bridge from local timber, and it has carried travelers for a hundred years.
  47. The photographer waited an hour for the clouds to part and still missed the perfect shot.
  48. Because the elevator was broken, we climbed eleven flights of stairs with all our luggage.
  49. The author rarely gives interviews, preferring to let the books speak for themselves.
  50. An old fisherman showed us how to mend the net using a wooden needle and waxed thread.
  51. The company promised faster delivery, but most customers noticed no difference at all.
  52. She planted an oak tree the day her son was born, and now it shades the entire yard.
  53. The children built a fortress from cardboard boxes and defended it until dinner.
  54. Our guide warned us that the trail becomes slippery whenever the fog settles in.
  55. The pharmacist double checked the prescription and explained how often to take each tablet.
  56. He whispered the answer to his friend, but the whole room heard him anyway.
  57. The festival begins with a parade through the old town and ends with fireworks over the river.
  58. After months of drought, the first heavy rain turned the dusty road into a stream.
  59. The carpenter measured the doorway three times before he made a single cut.
  60. I have always admired people who can stay calm when everything around them is falling apart.

Numbers, dates and symbols (60 sentences)

  1. The total comes to $1,250.50 after tax.
  2. The city has 3,482,000 residents.
  3. She was born in 1987 and moved here in 2004.
  4. The meeting is on 03/14/2025.
  5. The train leaves at 7:45 PM.
  6. Sales grew by 12.5% last quarter.
  7. Call us at 555-0142.
  8. Dr. Patel's office is at 40 Elm St.
  9. The package weighs 2.5 kg and is 30 cm wide.
  10. Send it to help@example.com.
  11. Visit www.example.org/start for details.
  12. He finished 3rd in the 21st race.
  13. It was -5°C this morning.
  14. NASA and the FBI issued a joint statement.
  15. Pi is roughly 3.14159.
  16. Tickets cost €45 each.
  17. Flight BA2490 departs from gate C14.
  18. The ratio is 3:1 and the pH is 7.4.
  19. The company raised $4.2M in 2023.
  20. He ran 26.2 miles in 3 hrs 15 min.
  21. Rent is $2,300 a month.
  22. A stamp costs $0.73.
  23. The fee is £18.99.
  24. The budget is $3.5 billion.
  25. She earns $85K a year.
  26. The deadline is Jan. 5, 2026.
  27. Music from the 1990s is back.
  28. It was built in the 18th century.
  29. The store opens at 09:30.
  30. We land at 12:05 a.m.
  31. The video is 1 hr 20 min long.
  32. Dial 1-800-555-0199 for support.
  33. The limit is 65 mph.
  34. Bake at 350°F for 25 min.
  35. The trail is 8.4 km long.
  36. Prices rose 3-5% this year.
  37. 5 + 7 = 12.
  38. Prof. Lee and Sen. Ortiz spoke first.
  39. Martin Luther King Jr. Blvd. is closed.
  40. Read chapter No. 9 tonight.
  41. They climbed Mt. Everest in May.
  42. Acme Inc. bought Zenith Ltd.
  43. J. K. Rowling wrote it.
  44. The CEO met the UN and the EU.
  45. Save the file as a PDF or a JPEG.
  46. Connect to Wi-Fi using WPA2.
  47. Update to version 2.0.1 today.
  48. It runs iOS 18 and Windows 11.
  49. Her birthday is the 22nd of June.
  50. The video has 1.2M views.
  51. About 15,000 people came.
  52. This is the 100th episode.
  53. The rate fell to 0.25%.
  54. Use promo code SAVE20 at checkout.
  55. The room number is 12B.
  56. He ruled from 1837 to 1901.
  57. Open 9 a.m. to 5 p.m. daily.
  58. Add 8 oz of butter.
  59. Winds reached 120 km/h.
  60. The phone has 256GB of storage.

Repeated words (20 lines)

  1. No, no, no, no, no. I said no.
  2. Row, row, row your boat, gently down the stream. Row, row, row your boat, gently down the stream.
  3. The code word is tango tango seven seven tango. I repeat: tango tango seven seven tango.
  4. Very, very, very good. Really, really, really, really good.
  5. Again and again and again and again and again, the waves hit the shore.
  6. She sells sea shells. She sells sea shells. She sells sea shells by the sea shore.
  7. Left, right, left, right, left, right, left. Keep marching, keep marching, keep marching.
  8. It was a long, long, long, long road, and a long, long, long way home.
  9. Go, go, go, go! Run, run, run!
  10. Please hold. Please hold. Please hold. Your call is important to us.
  11. Knock, knock, knock. Nobody answered. Knock, knock, knock. Still nobody answered.
  12. Yes, yes, yes, a thousand times yes.
  13. Never, never, never, never give up.
  14. Check one, check two. Check one, check two. Check one, check two.
  15. He said stop. She said stop. They all said stop. Nobody stopped.
  16. Down, down, down the rabbit hole she fell, down, down, down.
  17. Wait. Wait. Wait. Did you hear that? Wait.
  18. More and more and more people arrived, and more and more chairs were needed.
  19. Thank you, thank you, thank you all so much.
  20. Step by step by step by step, we climbed the tower.

Emotion control (10 sentences × neutral, angry, excited, scared, whispered)

  1. The package arrived at the front door this morning.
  2. I am going to the store, and I will be back at six.
  3. They told me the meeting has been moved to Thursday.
  4. There are two cars parked outside the building.
  5. She put the keys on the table next to the window.
  6. The report is due at the end of the month.
  7. We are having dinner at my sister's house tonight.
  8. He said the train leaves from platform four.
  9. The lights in the hallway are still on.
  10. Someone left a message for you at the front desk.

Other languages (8 sentences each)

  1. es El tren sale de la estación a primera hora de la mañana y llega a la costa antes del mediodía.
  2. es Mi abuela siempre decía que la paciencia es la madre de todas las ciencias.
  3. es Si mañana no llueve, iremos caminando hasta el mercado del pueblo.
  4. es Los niños jugaban en la plaza mientras sus padres tomaban café en la terraza.
  5. es Nunca había visto una ciudad tan tranquila a esas horas de la noche.
  6. es El médico me recomendó descansar y beber mucha agua durante toda la semana.
  7. es Compramos pan, queso y fruta para comer junto al río.
  8. es Cuando terminó la película, nadie se levantó de su asiento.
  9. fr Le boulanger ouvre sa boutique avant le lever du soleil, et tout le quartier sent le pain chaud.
  10. fr Nous avons marché longtemps le long de la rivière sans rencontrer personne.
  11. fr Elle a rangé ses livres, fermé la fenêtre et quitté la maison sans bruit.
  12. fr Les enfants attendaient le bus sous la pluie en chantant une vieille chanson.
  13. fr Mon frère travaille dans un petit restaurant près de la gare.
  14. fr Il faut tourner à gauche après le pont, puis continuer tout droit.
  15. fr Je me souviens encore du jardin de ma grand-mère et de ses grands arbres.
  16. fr Le train avait du retard, alors nous avons pris un café en attendant.
  17. de Der Zug fährt jeden Morgen pünktlich ab, aber heute hatte er eine halbe Stunde Verspätung.
  18. de Meine Schwester wohnt seit vielen Jahren in einer kleinen Stadt an der Küste.
  19. de Wenn das Wetter schön bleibt, gehen wir am Wochenende in den Bergen wandern.
  20. de Die Kinder spielten im Garten, während ihre Eltern das Abendessen vorbereiteten.
  21. de Ich habe meinen Schlüssel verloren und musste den Nachbarn um Hilfe bitten.
  22. de Der Arzt sagte, ich solle viel Wasser trinken und früh schlafen gehen.
  23. de Nach dem Regen roch die ganze Straße nach frischer Erde.
  24. de Wir haben den ganzen Abend über alte Zeiten gesprochen und viel gelacht.
  25. it Ogni domenica la famiglia si riunisce intorno al tavolo per un pranzo che dura tutto il pomeriggio.
  26. it Il vecchio pescatore conosceva il mare meglio di chiunque altro nel villaggio.
  27. it Abbiamo perso il treno, quindi siamo tornati a casa a piedi sotto la pioggia.
  28. it I bambini correvano nella piazza mentre le campane suonavano mezzogiorno.
  29. it Mia sorella lavora in una piccola libreria vicino alla stazione.
  30. it Dopo cena siamo usciti a fare una passeggiata lungo il fiume.
  31. it Non avevo mai visto un tramonto così bello sulla collina.
  32. it Il medico mi ha detto di riposare e di bere molta acqua.
  33. pt O mercado abre cedo, e os vendedores arrumam as frutas antes de o sol nascer.
  34. pt Ela guardou a carta na gaveta e nunca mais falou sobre o assunto.
  35. pt Se tivermos tempo, vamos visitar a aldeia onde os meus avós nasceram.
  36. pt As crianças brincavam na praia enquanto os pais conversavam à sombra.
  37. pt O meu irmão trabalha num pequeno café perto da estação.
  38. pt Depois do jantar, fomos caminhar ao longo do rio.
  39. pt Nunca tinha visto uma cidade tão calma àquela hora da noite.
  40. pt O médico disse que eu devia descansar e beber muita água.
  41. hi सुबह की ट्रेन समय पर आई और हम दोपहर से पहले शहर पहुँच गए।
  42. hi मेरी दादी हर शाम बच्चों को पुरानी कहानियाँ सुनाती थीं।
  43. hi अगर कल बारिश नहीं हुई तो हम बाज़ार तक पैदल जाएँगे।
  44. hi बच्चे पार्क में खेल रहे थे और उनके माता पिता चाय पी रहे थे।
  45. hi मेरा भाई स्टेशन के पास एक छोटी दुकान में काम करता है।
  46. hi डॉक्टर ने कहा कि मुझे आराम करना चाहिए और बहुत पानी पीना चाहिए।
  47. hi खाने के बाद हम नदी के किनारे टहलने गए।
  48. hi मैंने इतना सुंदर शहर पहले कभी नहीं देखा था।
  49. ja 朝の電車はいつも混んでいますが、今日は珍しく座ることができました。
  50. ja 祖母は毎年春になると、庭にたくさんの花を植えます。
  51. ja 雨が止んだら、川沿いの道を駅まで歩きましょう。
  52. ja 子供たちは公園で遊んでいて、両親はベンチで話していました。
  53. ja 兄は駅の近くの小さな店で働いています。
  54. ja 医者は、よく休んで水をたくさん飲むように言いました。
  55. ja 夕食の後、私たちは川のそばを散歩しました。
  56. ja こんなに美しい町は今まで見たことがありません。
  57. zh 早上的火车准时出发,我们在中午之前就到了海边。
  58. zh 奶奶每天晚上都会给孩子们讲一个老故事。
  59. zh 如果明天不下雨,我们就走路去镇上的市场。
  60. zh 孩子们在公园里玩,父母坐在长椅上聊天。
  61. zh 我哥哥在火车站附近的一家小店工作。
  62. zh 医生说我应该好好休息,多喝水。
  63. zh 晚饭以后,我们沿着河边散步。
  64. zh 我从来没有见过这么漂亮的城市。

Long text (one document, five lengths)

Lengths sent, in characters: 895, 1,875, 3,979, 7,938, 14,958. Each is the start of the same document, cut at the end of a sentence, and sent as a single request.

The full document (14,958 characters)
The lighthouse at the end of the point had not been lit for thirty years when Marta bought it. People in town assumed she would turn it into a guest house, and for a while she let them think so. In truth she had no plan at all. She had simply seen the photograph in a window, a white tower against a grey sea, and decided that she wanted to wake up there. The first winter was harder than she expected. Wind found its way through every window frame, the stove smoked, and the road flooded twice before the new year. She learned to keep a week of food in the pantry and a lamp beside the bed. She learned which stairs creaked and which ones held. By spring she knew the sound of every boat that passed. The fishermen began to wave, then to stop, then to leave a crab or two in a bucket by the door. One of them, an old man named Tobias, told her that his father had kept the light before the war. He showed her how the lens turned, how the oil was carried up, how a keeper could tell the weather from the color of the water at dusk. Marta wrote it all down in a blue notebook. On the first clear night of summer she climbed the tower with a borrowed lantern, set it in the middle of the old lens, and watched a thin beam swing slowly out across the bay. Down in the harbor, somebody sounded a horn in reply. Most people think of sleep as a single state, something you fall into at night and climb out of in the morning. In fact the brain moves through a repeating cycle of stages, and each one does different work. In the lightest stage, muscles relax and the heartbeat slows, but a small noise is enough to wake you. A deeper stage follows, in which the body repairs tissue and the brain clears away waste that built up during the day. This is the sleep that leaves you feeling rested, and it is the hardest to wake from. Later comes the stage in which most dreaming happens. The eyes move quickly under closed lids, breathing becomes uneven, and the brain is nearly as active as it is when you are awake. Researchers believe this stage helps to sort memories, keeping what matters and letting the rest fade. A full cycle takes about an hour and a half, and a healthy adult passes through four or five of them each night. The balance shifts as the night goes on. Early cycles contain more deep sleep, while later ones contain more dreaming. That is why cutting a night short does not simply remove a little of everything. It removes mostly dreaming sleep, and after several short nights in a row, people tend to feel forgetful and irritable even if they do not feel especially tired. Regular hours, a dark room, and a cool temperature all help the cycle run the way it should. Thank you for calling the support line. Before we begin, please have your account details ready, because the first thing our team will ask for is the name on the account and the address where the service is installed. If your connection has stopped working, start by looking at the small box near your front door. A steady green light means the line is fine, and the problem is probably inside the house. A blinking orange light means the box is trying to reconnect, and it is worth waiting a few minutes before doing anything else. If there is no light at all, check that the power cable has not been pulled loose. Next, find the router, which is the larger box with several cables in the back. Switch it off at the wall, count slowly to thirty, and switch it on again. It can take as long as five minutes to start up fully, so please be patient. If the connection returns, you do not need to do anything further. If it does not, stay on the line and one of our engineers will run a test from this end. They may ask you to plug a computer directly into the router with a cable, so it helps to have one nearby. Most faults are fixed during the first call. When a visit is needed, we will offer you the earliest morning or afternoon appointment and send a message the evening before to confirm the time. Bread needs only four things: flour, water, salt, and time. Begin the evening before you want to bake. Stir the flour and water together in a large bowl until no dry patches remain, cover it with a cloth, and leave it alone for half an hour. This rest lets the flour drink the water and makes the dough far easier to handle. Sprinkle the salt over the top, add the starter, and squeeze everything through your fingers until it feels smooth. For the next two hours, return to the bowl every thirty minutes. Wet your hand, reach under the dough, stretch it upward, and fold it over itself. Turn the bowl and repeat until you have gone all the way around. Each time, the dough will feel stronger and hold its shape a little longer. When it has grown by about half and small bubbles show along the edge, tip it gently onto the counter. Shape it into a tight round, place it upside down in a floured basket, and put it in the refrigerator overnight. In the morning, heat a heavy pot with its lid in a very hot oven. Turn the cold dough into the pot, cut a single line across the top with a sharp blade, and cover it. Bake until the loaf has risen, then remove the lid so the crust can darken. Let it cool completely on a rack before you cut it, however tempting the smell may be. A loaf cut too early will be gummy in the middle. Good evening, and here is the news at the top of the hour. City officials have approved a plan to rebuild the old harbor bridge, ending a debate that has lasted nearly a decade. Work will begin in the spring and is expected to continue for three years. During that time, traffic will be directed along the northern road, and a free ferry will carry passengers across the water every twenty minutes. Shop owners near the harbor say they are worried about losing customers, and the mayor has promised a fund to help the smallest businesses through the closure. In other news, schools across the region will open an hour later tomorrow because of the heavy snow forecast overnight. Parents are asked to check for messages before leaving home. The rail company has warned of delays on all lines, and drivers are advised to travel only if the journey is necessary. Turning to sport, the women's team has reached the national final for the first time in the club's history after a narrow win last night. The deciding goal came in the final minute, and thousands of supporters stayed in the stadium long after the whistle. The final will be played a week from Saturday. And finally, a retired teacher has returned a library book that her grandfather borrowed almost eighty years ago. The library says it will not be charging a fine, and the book is now on display beside the front desk. The old woman at the end of our street kept bees, and when I was nine she let me help. I remember the weight of the veil on my shoulders and the smell of the smoke she puffed across the hive to calm it. She never hurried. She would lift the lid as if she were opening a letter, set it down on the grass, and wait. The bees poured over the edge like dark water, and I held my breath. She told me that a hive has a mood, the same as a house does, and that you can hear it before you see it. A low, even hum means all is well. A high, sharp note means leave them alone and come back tomorrow. She showed me the queen, longer than the rest and always moving, with a ring of workers turning to face her wherever she went. She showed me the pale new wax and the darker comb where the honey was stored, each cell sealed with a thin white cap. That summer I was stung four times, and each time she pressed a cold spoon against my arm and said it was the price of the lesson. In the autumn she gave me a jar with my name written on the lid. I kept it on my windowsill for a year, unable to bring myself to open it, watching the light come through it gold in the mornings. I keep bees of my own now. I still do not hurry, and I still listen before I lift the lid. The old town is small enough to cross on foot in twenty minutes, and that is the best way to see it. Begin at the river gate, where the morning market fills the square with flowers, cheese, and baskets of bread. Follow the main street uphill past the clock tower. The clock has been wrong for as long as anyone can remember, and the people who live here have stopped noticing. Halfway up, a narrow lane on the left leads to a courtyard with a fountain and a single fig tree. Most visitors walk straight past it. Sit for a moment, because the benches are in the shade and the water is safe to drink. At the top of the hill stands the cathedral, which took two hundred years to finish and shows it. The lower walls are heavy and plain, while the upper windows are tall and full of colored glass. Climb the tower if your knees allow it. From the top you can see the whole valley, the vineyards on the southern slope, and on a clear day the mountains beyond. Come down by the eastern steps, which bring you to a row of small workshops where potters and weavers still sell what they make. End the walk at the bridge as the sun goes down. The stone turns the color of honey, the swallows come out over the water, and the cafes along the bank begin to light their lamps. For most of history, a message could travel no faster than the person carrying it. A letter crossed a country at the speed of a horse and crossed an ocean at the speed of a ship. News of a battle might arrive weeks after the war had ended. People planned their lives around that delay without thinking of it as a delay at all. Then, within a single lifetime, everything changed. Wires were strung between cities, and a clerk in one town could tap out a sentence that was read in another town a moment later. At first the service was expensive, and people counted every word. Businesses invented short codes so that a single word could stand for an entire instruction. Newspapers began to print reports from distant places on the day they happened. Railways used the wires to keep trains from meeting on the same track. Families learned to dread the knock of a messenger, because urgent news was so often bad news. Not everyone welcomed the change. Some writers complained that the world had become too hurried and that nobody had time to think before replying. Others worried that false reports would spread before anyone could correct them. Those complaints sound familiar today. Every new way of sending words faster has brought the same hopes and the same fears, and each time people have adjusted, slowly, until the new speed felt ordinary. Before you use your new kettle for the first time, remove all the packaging and wash the inside with warm water. Do not put the base in water or allow it to get wet. Place the base on a flat, dry surface away from the edge of the counter, and keep the cord where it cannot be pulled by a child. Fill the kettle with fresh water to somewhere between the lowest and highest marks on the side. Too little water can damage the heating plate, and too much may cause boiling water to spill from the spout. Close the lid firmly until you hear it click. Set the kettle on the base and press the switch down. A light will come on to show that it is heating. When the water boils, the kettle switches itself off and the light goes out. Wait a few seconds before you lift it, and always pour slowly. Never open the lid while steam is rising. If you want to boil the water again, let the kettle rest for half a minute first. Over time, a pale crust may form on the inside. This is harmless, but it makes the kettle slower and louder. To remove it, fill the kettle halfway with equal parts water and white vinegar, bring it to the boil, and leave it to stand for an hour. Pour the mixture away, rinse several times, and boil a full kettle of clean water before you make a drink. At low tide the shore becomes a different country. The water draws back and leaves a wide floor of rock and weed, shining and loud with small sounds. Crabs click across the stones. Water drains through the gravel with a long, soft hiss. In every hollow there is a pool, and every pool is a world. Kneel beside one and keep still, and after a minute it comes to life. A shrimp, clear as glass, drifts out from under a ledge. A snail drags its shell across a leaf of weed, leaving a clean path behind it. Something that looked like a red flower opens its arms and begins to feed. The animals that live here have solved a hard problem. Twice a day the sea covers them, cold and rough and full of food. Twice a day it leaves them to the sun, the wind, the rain, and the birds. Some close themselves tight and wait. Some hide in wet cracks where the air stays cool. Some simply hold on. The limpet returns to the same spot on the same rock every time the tide falls, and over the years it wears a shallow ring there that fits its shell exactly. When the water turns, it comes back quietly at first, filling the lowest pools and joining them together. Then the first real wave breaks over the outer rocks, and it is time to pick up your shoes and walk back to the sand. The letter arrived on a Tuesday, in an envelope so thin that Daniel could read the first line through the paper. He did not open it right away. He set it on the kitchen table, made coffee, fed the cat, and stood at the window watching the bus pull away from the stop without him. His sister called twice and he let the phone ring. For eleven years he had worked in the same office, at the same desk, beside the same window that did not open. He knew which floorboard creaked outside the manager's door and which drawer held the good pens. He had always assumed that leaving would be his own decision, made on some bright morning when he finally felt ready. He had not expected a letter. When he did open it, the words were polite and brief. They thanked him for his service and wished him well. He read it three times, folded it along the same creases, and put it in his coat pocket. Then he did something he had not done in years. He took the long way into town, through the park, past the pond where he had learned to skate as a boy. The ice was thin and grey at the edges. A man was selling roasted chestnuts from a cart, and Daniel bought a bag and ate them on a bench, burning his fingers. By the time the bag was empty, he had begun, cautiously, to make a list of things he might do instead. Saving money is simple to describe and difficult to do. The idea is to spend a little less than you earn and to put the difference somewhere you will not touch it. The difficulty is that the reward lies far in the future, while the cost is felt today. A useful first step is to write down everything you spend for one month. Most people are surprised by the result. Small, regular purchases add up to more than the large ones that we remember and worry about. Once you can see where the money goes, choose one or two habits to change rather than trying to change everything at once.

Music models tested

ModelAPI modelReached throughSingsControls used
Stable Audio 2.5
Stability AI
stable-audio-2.5ReplicatenoLength field
Eleven Music
ElevenLabs
music_v1ReplicateyesLength field, instrumental switch
ACE-Step 1.5
ACE Studio / StepFun (open source)
ace-step-1.5ReplicateyesLyrics, tempo and length fields
Lyria 3 Clip
Google
lyria-3-clip-previewGemini APIyesPrompt only (fixed ~30 s clips)
Lyria 3.5
Google
lyria-3.5Gemini APIyesPrompt only
Lyria 3 Pro
Google
lyria-3-pro-previewGemini APIyesPrompt only
MusicGen Large (stereo)
Meta (open source)
musicgen-stereo-largeReplicatenoLength field
MiniMax Music 2.6
MiniMax
music-2.6ReplicateyesLyrics field, instrumental switch

Not tested yet: Suno (no public api), Udio (no public api). Every model gets the same prompt text; where a model has dedicated fields for lyrics, tempo, length or "instrumental", those are filled in as well.

Music tests and scoring

Lyric accuracy 30% of overall

Test: 12 short songs across styles from pop and rock to hip hop, reggae and a lullaby, each with a verse and chorus we wrote, about 55 words.
Scoring: The vocal stem is separated from the mix, and both the stem and the full mix are transcribed by two speech recognisers. Score = share of the supplied lyric words that can be heard in order, taken from the best transcript. Repeats and extra words do not lower the score; they are reported separately.

Genre 20% of overall

Test: 24 instrumental briefs across 12 genres: lo-fi hip hop, country, funk, techno, punk rock, drum and bass, reggae, disco, heavy metal, jazz, bossa nova and trap.
Scoring: Two AI listeners each pick one label from a fixed list of 16 genres without seeing the prompt. A clip passes when either names the requested genre. Picking the right one of 16 by accident is unlikely, so a pass is strong evidence; a fail can also be a listener error.

Tempo 20% of overall

Test: The same 24 briefs. Each genre is asked for at two different tempos, from 75 to 180 BPM, so a model cannot pass by playing the genre at its usual speed.
Scoring: We measure the beat grid directly from the audio. A clip passes when the grid at the requested tempo, or its half or double, is at least 90% as strong as the strongest grid found at any tempo. Half and double time count because the two are musically the same pulse.

Length 10% of overall

Test: The same 24 briefs, each asking for a 30 second track. Models with a length setting have it set to 30.
Scoring: Pass when the track is within 4.5 seconds of 30.

No vocals when asked 10% of overall

Test: The same 24 briefs, all marked instrumental.
Scoring: The vocal stem is separated. A clip fails only when that stem is loud and contains at least 8 recognisable sung words.

Clean ending 10% of overall

Test: Every track in the benchmark.
Scoring: Pass when the last 0.4 seconds are at least 12 dB quieter than the track as a whole (a fade or a final note). A track that is cut off mid-phrase fails.

Speed not in overall

Test: The 24 instrumental briefs.
Scoring: Median seconds to generate. Direct calls are full request time; Replicate-hosted models use the compute time Replicate reports. Not part of the overall score.

Instrumental briefs

  1. Lo-fi hip hop beat with mellow electric piano, soft drums and vinyl crackle. 80 BPM. Instrumental only, no vocals. 30 seconds long.
  2. Country tune with strummed acoustic guitar, fiddle and brushed drums. 95 BPM. Instrumental only, no vocals. 30 seconds long.
  3. Funk groove with slap bass, tight rhythm guitar and a horn section. 110 BPM. Instrumental only, no vocals. 30 seconds long.
  4. Techno track with a four-on-the-floor kick drum and an acid bassline. 130 BPM. Instrumental only, no vocals. 30 seconds long.
  5. Punk rock with distorted power chords and fast, energetic drums. 160 BPM. Instrumental only, no vocals. 30 seconds long.
  6. Drum and bass with fast breakbeats and a deep sub bass. 174 BPM. Instrumental only, no vocals. 30 seconds long.
  7. Lo-fi hip hop beat with a dusty jazz guitar loop, soft kick and tape hiss. 92 BPM. Instrumental only, no vocals. 30 seconds long.
  8. Upbeat country tune with banjo, pedal steel guitar and a train-beat snare. 120 BPM. Instrumental only, no vocals. 30 seconds long.
  9. Slow funk groove with clavinet, wah guitar and a syncopated bass line. 98 BPM. Instrumental only, no vocals. 30 seconds long.
  10. Hard techno track with a pounding kick drum, hi-hats and dark synth stabs. 140 BPM. Instrumental only, no vocals. 30 seconds long.
  11. Fast punk rock with buzzing guitars, driving bass and crashing cymbals. 180 BPM. Instrumental only, no vocals. 30 seconds long.
  12. Liquid drum and bass with rolling breakbeats, warm pads and a smooth sub bass. 165 BPM. Instrumental only, no vocals. 30 seconds long.
  13. Reggae groove with offbeat guitar chops, a heavy bass line and a one-drop drum beat. 75 BPM. Instrumental only, no vocals. 30 seconds long.
  14. Bright reggae track with organ bubble, offbeat guitar and a steady rimshot. 90 BPM. Instrumental only, no vocals. 30 seconds long.
  15. Disco track with a four-on-the-floor beat, octave bass, string stabs and open hi-hats. 120 BPM. Instrumental only, no vocals. 30 seconds long.
  16. Laid-back disco groove with rhythm guitar, electric piano and hand claps. 110 BPM. Instrumental only, no vocals. 30 seconds long.
  17. Heavy metal with palm-muted distorted guitar riffs and double bass drums. 150 BPM. Instrumental only, no vocals. 30 seconds long.
  18. Mid-tempo heavy metal with a crushing guitar riff, bass and pounding drums. 120 BPM. Instrumental only, no vocals. 30 seconds long.
  19. Swing jazz with walking upright bass, ride cymbal, piano and a saxophone lead. 140 BPM. Instrumental only, no vocals. 30 seconds long.
  20. Relaxed jazz trio with brushed drums, upright bass and piano. 100 BPM. Instrumental only, no vocals. 30 seconds long.
  21. Bossa nova with nylon-string guitar, soft shaker and a gentle rim click. 100 BPM. Instrumental only, no vocals. 30 seconds long.
  22. Lively bossa nova with nylon-string guitar, flute and light percussion. 125 BPM. Instrumental only, no vocals. 30 seconds long.
  23. Trap beat with booming 808 bass, rapid hi-hat rolls and a dark bell melody. 150 BPM. Instrumental only, no vocals. 30 seconds long.
  24. Trap beat with deep 808 bass, snappy snare and a moody piano loop. 130 BPM. Instrumental only, no vocals. 30 seconds long.

Genre labels offered to the AI listeners: lo-fi hip hop, country, funk, techno, punk rock, drum and bass, reggae, disco, heavy metal, jazz, bossa nova, trap, classical, ambient, folk, blues.

Songs with lyrics

Upbeat pop song with a clear female lead vocal, bright synths and a steady beat
[Verse] Morning light is on the window Coffee steam is in the air I have got a pocket full of maybe And a day that does not care [Chorus] So I am running to the river With the summer in my hands Every road is getting wider Every stranger is a friend
Driving rock song with a clear male lead vocal, electric guitars and live drums
[Verse] The engine coughs, the highway hums A hundred miles of dust and sun I left my name in a border town And I am not turning back around [Chorus] Burn the map and hold the wheel Nothing matters but what is real Tonight the thunder knows my name And I will never be the same
Gentle acoustic folk song with a clear lead vocal and fingerpicked guitar
[Verse] My father built this wooden gate With iron nails and patient hands The paint has faded, the hinges ache But still it swings and still it stands [Chorus] Carry me home along the valley Where the apple trees are tall Carry me home before the winter Lays its silver on the wall
Slow soul ballad with a clear, warm lead vocal, electric piano and soft drums
[Verse] Midnight in the kitchen, the radio is low You are dancing in your socks across the floor I could watch you until the morning comes And still be asking for a little more [Chorus] Stay a while, stay a while Let the city fall asleep Stay a while, stay a while You are everything I need
Country song with a clear male lead vocal, acoustic guitar and pedal steel
[Verse] There is a porch light burning on a gravel road A screen door singing in the evening breeze My mother waving with a dish towel in her hand And fireflies rising from the cedar trees [Chorus] Take me back to the county line Where the creek runs cold and clear Take me back to that old front yard I left my heart right here
Hip hop track with a clear rapped male vocal over a boom bap drum beat and a piano loop
[Verse] Woke up early with a plan in my pocket Shoes by the door and the bus on the way Counting every dollar, every minute, every promise Building something better than yesterday [Chorus] Keep it moving, keep it steady One more step and I am ready Keep it moving, keep it steady Nothing in the world can stop me
Reggae song with a clear relaxed lead vocal, offbeat guitar and a deep bass line
[Verse] Sun is rising over the water Fishing boats are heading out to sea I have no worry on my shoulder Only the morning and the breeze [Chorus] Easy now, easy now Let the trouble drift away Easy now, easy now Tomorrow is another day
Disco song with a clear female lead vocal, four-on-the-floor beat, strings and funky bass
[Verse] Mirror ball is turning in the ceiling Silver shoes are sliding on the floor Somebody is calling out my number And I cannot sit still anymore [Chorus] Dance with me until the lights come on Spin me round and never let me go Dance with me until the night is gone Tonight is all we need to know
Slow piano ballad with a clear, emotional female lead vocal and soft strings
[Verse] I found your letter in a winter coat The ink had faded but the words were clear You wrote that love is like a paper boat It only travels if the wind is near [Chorus] So here I am beside the water Waiting for the wind to rise Here I am beside the water With your letter and the sky
Indie rock song with a clear male lead vocal, jangly guitars and a steady drum beat
[Verse] We painted all the bedroom walls in yellow And hung a paper lantern from the light You said the city sounded like a cello Playing slowly in the middle of the night [Chorus] We were young and we were certain We were loud and we were kind Open up the kitchen curtain Leave the quiet years behind
Electronic dance pop song with a clear female lead vocal, pulsing synths and a strong kick drum
[Verse] Neon rivers running through the station Every heartbeat matching with the sound I can feel a brand new conversation Rising from the speakers in the ground [Chorus] Turn it up and let it carry me Higher than the rooftops and the rain Turn it up and let it carry me I will never come back down again
Soft lullaby with a clear, gentle female lead vocal, music box and quiet acoustic guitar
[Verse] The moon is a lantern above the hill The owls are counting the sheep The river is humming a song so still It is time for my darling to sleep [Chorus] Close your eyes, little one The stars will keep you warm Close your eyes, little one I will hold you until dawn

Known limits

What we tried and dropped

Conflict of interest, and our own rows

AnyToSpeech is a product that sells access to some of the models tested here, including voices from Google, Inworld and OpenAI, and music from Google. No vendor paid for, saw or influenced these results.

The speech leaderboard also includes AnyToSpeech itself, marked "ours", as two rows: Standard voices and Premium and Characters. AnyToSpeech does not make a voice model. Standard voices run on Gemini 3.8 Flash TTS and Premium and Character voices run on Inworld TTS-2; both go through the site's own pipeline, which splits long text at sentence ends and joins the pieces, adds number-reading hints, detects the language, and retries or falls back when a piece fails. The pipeline was called exactly as the site calls it, with no prompt, voice or setting chosen for this test, and it is scored by the same code as every other row. Because it splits and retries internally, its long-text and reliability results describe the product, not a single model request.

The defence against bias is the method: fixed inputs, measured outputs, published rules, and audio you can check.

What is coming

To have a model added, or to dispute a result, contact us.

Think a score is wrong? Check it.

Every test item, scoring rule and known limit is published, and the audio behind each number is on the page.

Read the methodology