Spaces:
Running on Zero
Running on Zero
Add five Chinese and English poetry speech recordings
Browse files- .gitattributes +5 -0
- examples/poetry/README.md +51 -0
- examples/poetry/manifest.json +91 -0
- examples/poetry/prompts.json +42 -0
- examples/poetry/quiet-night.wav +3 -0
- examples/poetry/sonnet-18.wav +3 -0
- examples/poetry/spring-dawn.wav +3 -0
- examples/poetry/stork-tower.wav +3 -0
- examples/poetry/the-tyger.wav +3 -0
.gitattributes
CHANGED
|
@@ -33,3 +33,8 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
+
examples/poetry/quiet-night.wav filter=lfs diff=lfs merge=lfs -text
|
| 37 |
+
examples/poetry/sonnet-18.wav filter=lfs diff=lfs merge=lfs -text
|
| 38 |
+
examples/poetry/spring-dawn.wav filter=lfs diff=lfs merge=lfs -text
|
| 39 |
+
examples/poetry/stork-tower.wav filter=lfs diff=lfs merge=lfs -text
|
| 40 |
+
examples/poetry/the-tyger.wav filter=lfs diff=lfs merge=lfs -text
|
examples/poetry/README.md
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Poetry voice examples
|
| 2 |
+
|
| 3 |
+
Five public-domain poetry readings synthesized with Qwen3-TTS 1.7B VoiceDesign. Chinese poems are complete; English readings are excerpts of the original English poems. No modern translations are used.
|
| 4 |
+
|
| 5 |
+
Select a poem in the application to load its recording into **Audio → MIDI**. You can listen immediately and then generate piano MIDI. These assets contain speech, not precomputed piano reconstructions.
|
| 6 |
+
|
| 7 |
+
| Poem | Author | Language | Duration | Audio |
|
| 8 |
+
|---|---|---|---:|---|
|
| 9 |
+
| 静夜思 | 李白 | Chinese | 8.64 s | [Listen](quiet-night.wav) |
|
| 10 |
+
| 春晓 | 孟浩然 | Chinese | 8.16 s | [Listen](spring-dawn.wav) |
|
| 11 |
+
| 登鹳雀楼 | 王之涣 | Chinese | 7.60 s | [Listen](stork-tower.wav) |
|
| 12 |
+
| Sonnet 18 (opening couplet) | William Shakespeare | English | 4.64 s | [Listen](sonnet-18.wav) |
|
| 13 |
+
| The Tyger (first stanza) | William Blake | English | 8.24 s | [Listen](the-tyger.wav) |
|
| 14 |
+
|
| 15 |
+
All recordings use the requested voice description:
|
| 16 |
+
|
| 17 |
+
> 一位温柔自信的年轻女性,声音清晰,语气亲切,表达轻快而富有感染力。
|
| 18 |
+
|
| 19 |
+
Model revision, seed, generation settings, timings, and file hashes are recorded in [manifest.json](manifest.json); [prompts.json](prompts.json) contains exact texts. Audio is 24 kHz mono. These fixtures were generated on CUDA with bfloat16 and seed 42 reset for every sample; the application also retains its CPU inference path.
|
| 20 |
+
|
| 21 |
+
## Poems
|
| 22 |
+
|
| 23 |
+
### 静夜思 — 李白
|
| 24 |
+
|
| 25 |
+
床前明月光,疑是地上霜。举头望明月,低头思故乡。
|
| 26 |
+
|
| 27 |
+
### 春晓 — 孟浩然
|
| 28 |
+
|
| 29 |
+
春眠不觉晓,处处闻啼鸟。夜来风雨声,花落知多少。
|
| 30 |
+
|
| 31 |
+
### 登鹳雀楼 — 王之涣
|
| 32 |
+
|
| 33 |
+
白日依山尽,黄河入海流。欲穷千里目,更上一层楼。
|
| 34 |
+
|
| 35 |
+
### Sonnet 18 (opening couplet) — William Shakespeare
|
| 36 |
+
|
| 37 |
+
Shall I compare thee to a summer’s day? Thou art more lovely and more temperate.
|
| 38 |
+
|
| 39 |
+
### The Tyger (first stanza) — William Blake
|
| 40 |
+
|
| 41 |
+
Tyger Tyger, burning bright, In the forests of the night; What immortal hand or eye, Could frame thy fearful symmetry?
|
| 42 |
+
|
| 43 |
+
## Regenerate
|
| 44 |
+
|
| 45 |
+
```bash
|
| 46 |
+
python scripts/generate_speech.py --device cpu \
|
| 47 |
+
--model Qwen/Qwen3-TTS-12Hz-1.7B-VoiceDesign \
|
| 48 |
+
--samples examples/poetry/prompts.json --out outputs/poetry
|
| 49 |
+
```
|
| 50 |
+
|
| 51 |
+
Different hardware and library versions can change the generated voice. Use the committed recordings for direct fitting comparisons.
|
examples/poetry/manifest.json
ADDED
|
@@ -0,0 +1,91 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "Qwen/Qwen3-TTS-12Hz-1.7B-VoiceDesign",
|
| 3 |
+
"source": {
|
| 4 |
+
"repo": "Qwen/Qwen3-TTS-12Hz-1.7B-VoiceDesign",
|
| 5 |
+
"revision": "5ecdb67327fd37bb2e042aab12ff7391903235d3"
|
| 6 |
+
},
|
| 7 |
+
"mode": "voice_design",
|
| 8 |
+
"seed": 42,
|
| 9 |
+
"seed_policy": "Reset RNG before each sample",
|
| 10 |
+
"device": "cuda",
|
| 11 |
+
"dtype": "bfloat16",
|
| 12 |
+
"attention": "sdpa",
|
| 13 |
+
"max_new_tokens": 512,
|
| 14 |
+
"samples": [
|
| 15 |
+
{
|
| 16 |
+
"name": "quiet-night",
|
| 17 |
+
"title": "静夜思",
|
| 18 |
+
"author": "李白",
|
| 19 |
+
"text": "床前明月光,疑是地上霜。举头望明月,低头思故乡。",
|
| 20 |
+
"language": "Chinese",
|
| 21 |
+
"instruct": "一位温柔自信的年轻女性,声音清晰,语气亲切,表达轻快而富有感染力。",
|
| 22 |
+
"seed": 42,
|
| 23 |
+
"sample_rate": 24000,
|
| 24 |
+
"seconds": 8.64,
|
| 25 |
+
"wall_seconds": 24.568893432617188,
|
| 26 |
+
"file": "quiet-night.wav",
|
| 27 |
+
"sha256": "7bdbca39bd9c406b639b9fda231219484711ca319bd4da9c0c9d991874707981",
|
| 28 |
+
"bytes": 829520
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"name": "spring-dawn",
|
| 32 |
+
"title": "春晓",
|
| 33 |
+
"author": "孟浩然",
|
| 34 |
+
"text": "春眠不觉晓,处处闻啼鸟。夜来风雨声,花落知多少。",
|
| 35 |
+
"language": "Chinese",
|
| 36 |
+
"instruct": "一位温柔自信的年轻女性,声音清晰,语气亲切,表达轻快而富有感染力。",
|
| 37 |
+
"seed": 42,
|
| 38 |
+
"sample_rate": 24000,
|
| 39 |
+
"seconds": 8.16,
|
| 40 |
+
"wall_seconds": 22.90131402015686,
|
| 41 |
+
"file": "spring-dawn.wav",
|
| 42 |
+
"sha256": "316f58a7865d1f31822471e2e84f507af683ece80dab7b1f41414310796b6b67",
|
| 43 |
+
"bytes": 783440
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"name": "stork-tower",
|
| 47 |
+
"title": "登鹳雀楼",
|
| 48 |
+
"author": "王之涣",
|
| 49 |
+
"text": "白日依山尽,黄河入海流。欲穷千里目,更上一层楼。",
|
| 50 |
+
"language": "Chinese",
|
| 51 |
+
"instruct": "一位温柔自信的年轻女性,声音清晰,语气亲切,表达轻快而富有感染力。",
|
| 52 |
+
"seed": 42,
|
| 53 |
+
"sample_rate": 24000,
|
| 54 |
+
"seconds": 7.6,
|
| 55 |
+
"wall_seconds": 21.23459029197693,
|
| 56 |
+
"file": "stork-tower.wav",
|
| 57 |
+
"sha256": "0aae65399547fef6d999e4b929b252fc9a428e82c7f159779d3a9bd1ca061c08",
|
| 58 |
+
"bytes": 729680
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"name": "sonnet-18",
|
| 62 |
+
"title": "Sonnet 18 (opening couplet)",
|
| 63 |
+
"author": "William Shakespeare",
|
| 64 |
+
"text": "Shall I compare thee to a summer’s day? Thou art more lovely and more temperate.",
|
| 65 |
+
"language": "English",
|
| 66 |
+
"instruct": "一位温柔自信的年轻女性,声音清晰,语气亲切,表达轻快而富有感染力。",
|
| 67 |
+
"seed": 42,
|
| 68 |
+
"sample_rate": 24000,
|
| 69 |
+
"seconds": 4.64,
|
| 70 |
+
"wall_seconds": 14.138381242752075,
|
| 71 |
+
"file": "sonnet-18.wav",
|
| 72 |
+
"sha256": "55f718b8da437c778f35c9a22c8b0b8c7ab8c176e8a91ad460f036afa53cc5e5",
|
| 73 |
+
"bytes": 445520
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"name": "the-tyger",
|
| 77 |
+
"title": "The Tyger (first stanza)",
|
| 78 |
+
"author": "William Blake",
|
| 79 |
+
"text": "Tyger Tyger, burning bright, In the forests of the night; What immortal hand or eye, Could frame thy fearful symmetry?",
|
| 80 |
+
"language": "English",
|
| 81 |
+
"instruct": "一位温柔自信的年轻女性,声音清晰,语气亲切,表达轻快而富有感染力。",
|
| 82 |
+
"seed": 42,
|
| 83 |
+
"sample_rate": 24000,
|
| 84 |
+
"seconds": 8.24,
|
| 85 |
+
"wall_seconds": 24.128756523132324,
|
| 86 |
+
"file": "the-tyger.wav",
|
| 87 |
+
"sha256": "bd807433cebced5629a63b5cf0ecaf213c301ccf3fc110b9b20ab74dc6f94d04",
|
| 88 |
+
"bytes": 791120
|
| 89 |
+
}
|
| 90 |
+
]
|
| 91 |
+
}
|
examples/poetry/prompts.json
ADDED
|
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"name": "quiet-night",
|
| 4 |
+
"title": "静夜思",
|
| 5 |
+
"author": "李白",
|
| 6 |
+
"text": "床前明月光,疑是地上霜。举头望明月,低头思故乡。",
|
| 7 |
+
"language": "Chinese",
|
| 8 |
+
"instruct": "一位温柔自信的年轻女性,声音清晰,语气亲切,表达轻快而富有感染力。"
|
| 9 |
+
},
|
| 10 |
+
{
|
| 11 |
+
"name": "spring-dawn",
|
| 12 |
+
"title": "春晓",
|
| 13 |
+
"author": "孟浩然",
|
| 14 |
+
"text": "春眠不觉晓,处处闻啼鸟。夜来风雨声,花落知多少。",
|
| 15 |
+
"language": "Chinese",
|
| 16 |
+
"instruct": "一位温柔自信的年轻女性,声音清晰,语气亲切,表达轻快而富有感染力。"
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"name": "stork-tower",
|
| 20 |
+
"title": "登鹳雀楼",
|
| 21 |
+
"author": "王之涣",
|
| 22 |
+
"text": "白日依山尽,黄河入海流。欲穷千里目,更上一层楼。",
|
| 23 |
+
"language": "Chinese",
|
| 24 |
+
"instruct": "一位温柔自信的年轻女性,声音清晰,语气亲切,表达轻快而富有感染力。"
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"name": "sonnet-18",
|
| 28 |
+
"title": "Sonnet 18 (opening couplet)",
|
| 29 |
+
"author": "William Shakespeare",
|
| 30 |
+
"text": "Shall I compare thee to a summer’s day? Thou art more lovely and more temperate.",
|
| 31 |
+
"language": "English",
|
| 32 |
+
"instruct": "一位温柔自信的年轻女性,声音清晰,语气亲切,表达轻快而富有感染力。"
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"name": "the-tyger",
|
| 36 |
+
"title": "The Tyger (first stanza)",
|
| 37 |
+
"author": "William Blake",
|
| 38 |
+
"text": "Tyger Tyger, burning bright, In the forests of the night; What immortal hand or eye, Could frame thy fearful symmetry?",
|
| 39 |
+
"language": "English",
|
| 40 |
+
"instruct": "一位温柔自信的年轻女性,声音清晰,语气亲切,表达轻快而富有感染力。"
|
| 41 |
+
}
|
| 42 |
+
]
|
examples/poetry/quiet-night.wav
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7bdbca39bd9c406b639b9fda231219484711ca319bd4da9c0c9d991874707981
|
| 3 |
+
size 829520
|
examples/poetry/sonnet-18.wav
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:55f718b8da437c778f35c9a22c8b0b8c7ab8c176e8a91ad460f036afa53cc5e5
|
| 3 |
+
size 445520
|
examples/poetry/spring-dawn.wav
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:316f58a7865d1f31822471e2e84f507af683ece80dab7b1f41414310796b6b67
|
| 3 |
+
size 783440
|
examples/poetry/stork-tower.wav
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0aae65399547fef6d999e4b929b252fc9a428e82c7f159779d3a9bd1ca061c08
|
| 3 |
+
size 729680
|
examples/poetry/the-tyger.wav
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bd807433cebced5629a63b5cf0ecaf213c301ccf3fc110b9b20ab74dc6f94d04
|
| 3 |
+
size 791120
|