From 3efb517b773fbf052adcd1186ced91ba1a9977eb Mon Sep 17 00:00:00 2001 From: Brandon Thomas Date: Wed, 1 Jul 2020 22:56:06 -0400 Subject: [PATCH] use new melgan for speakers --- tts_service/models.toml | 33 ++++++++++++++++++--------------- 1 file changed, 18 insertions(+), 15 deletions(-) diff --git a/tts_service/models.toml b/tts_service/models.toml index 47f1d61980..4edf790c8c 100644 --- a/tts_service/models.toml +++ b/tts_service/models.toml @@ -14,21 +14,21 @@ name = "Bill Clinton" slug = "bill-clinton" model_pipeline = "ArpabetGlowTtsMelgan" glow_tts = "glow-tts_bill_clinton_ljstx_2020.06.28_expr-1_chkpt-1659.torchjit" -melgan = "melgan_manyvoice1.0_2020-06-28_12d5838_6730.torchjit" +melgan = "manyvoice1.0_2020-06-28_12d5838_7050.torchjit" [[speakers]] name = "Christopher Lee" slug = "christopher-lee" model_pipeline = "ArpabetGlowTtsMelgan" glow_tts = "glow-tts_christopher_lee_ljstx_2020.06.28_expr-1_chkpt-4100.torchjit" -melgan = "melgan_manyvoice1.0_2020-06-28_12d5838_6730.torchjit" +melgan = "manyvoice1.0_2020-06-28_12d5838_7050.torchjit" [[speakers]] -name = "Barack Obama" +name = "Barack Obama (needs work)" slug = "barack-obama" model_pipeline = "ArpabetGlowTtsMelgan" glow_tts = "glow-tts_barack_obama_ljstx_2020.06.28_expr-1_chkpt-1729.torchjit" -melgan = "melgan_manyvoice1.0_2020-06-28_12d5838_6730.torchjit" +melgan = "manyvoice1.0_2020-06-28_12d5838_7050.torchjit" sample_rate_hz = 22000 # This is lower than normal [[speakers]] @@ -36,81 +36,84 @@ name = "Betty White" slug = "betty-white" model_pipeline = "ArpabetGlowTtsMelgan" glow_tts = "glow-tts_betty_white_ljstx_2020.06.28_expr-1_chkpt-1666.torchjit" -melgan = "melgan_manyvoice1.0_2020-06-28_12d5838_6730.torchjit" +melgan = "manyvoice1.0_2020-06-28_12d5838_7050.torchjit" [[speakers]] name = "Dr. Phil McGraw" slug = "dr-phil-mcgraw" model_pipeline = "ArpabetGlowTtsMelgan" glow_tts = "glow-tts_dr_phil_mcgraw_ljstx_2020.06.28_expr-1_chkpt-1813.torchjit" -melgan = "melgan_manyvoice1.0_2020-06-28_12d5838_6730.torchjit" +melgan = "manyvoice1.0_2020-06-28_12d5838_7050.torchjit" [[speakers]] name = "Fred Rogers" slug = "fred-rogers" model_pipeline = "ArpabetGlowTtsMelgan" glow_tts = "glow-tts_fred_rogers_ljstx_2020.06.28_expr-1_chkpt-2575.torchjit" -melgan = "melgan_manyvoice1.0_2020-06-28_12d5838_6730.torchjit" +melgan = "manyvoice1.0_2020-06-28_12d5838_7050.torchjit" [[speakers]] name = "Danny Devito" slug = "danny-devito" model_pipeline = "ArpabetGlowTtsMelgan" glow_tts = "glow-tts_danny_devito_ljstx_2020.06.28_expr-1_chkpt-3757.torchjit" -melgan = "melgan_manyvoice1.0_2020-06-28_12d5838_6730.torchjit" +melgan = "manyvoice1.0_2020-06-28_12d5838_7050.torchjit" [[speakers]] name = "Donald Trump" slug = "donald-trump" model_pipeline = "ArpabetGlowTtsMelgan" -melgan = "melgan_trump_txlearn_2020-05-30_steno_chkpt9800.torchjit" glow_tts = "glow_tts_ljs_txlearn_arpabet_2020-06-01_G_1689-2.torchjit" +#melgan = "manyvoice1.0_2020-06-28_12d5838_7050.torchjit" +melgan = "melgan_trump_txlearn_2020-05-30_steno_chkpt9800.torchjit" +sample_rate_hz = 21000 # This is lower than normal [[speakers]] name = "Gilbert Gottfried" slug = "gilbert-gottfried" model_pipeline = "ArpabetGlowTtsMelgan" -melgan = "manyvoice1.0_2020-06-28_12d5838_6610.torchjit" glow_tts = "glow-tts_gilbert-gottfried_ljstx_2020.06.28_expr-1_chkpt-1766.torchjit" +melgan = "manyvoice1.0_2020-06-28_12d5838_7050.torchjit" [[speakers]] name = "Ronald Reagan" slug = "ronald-reagan" model_pipeline = "ArpabetGlowTtsMelgan" -melgan = "melgan_vocodes1.0_multispeaker_2020.06.19_12d5838_experiment-2_chkpt-10400.torchjit" glow_tts = "glow-tts_solo_reagan_ljstx_2020.06.22_expr-1_chkpt-1406.torchjit" +melgan = "manyvoice1.0_2020-06-28_12d5838_7050.torchjit" sample_rate_hz = 21000 # This is lower than normal [[speakers]] name = "John Oliver" slug = "john-oliver" model_pipeline = "ArpabetGlowTtsMelgan" -melgan = "melgan_vocodes1.0_multispeaker_2020.06.19_12d5838_experiment-2_chkpt-10400.torchjit" glow_tts = "glow-tts_solo_oliver_ljstx_2020.06.22_expr-1_chkpt-1189.torchjit" +melgan = "manyvoice1.0_2020-06-28_12d5838_7050.torchjit" sample_rate_hz = 21000 # This is lower than normal [[speakers]] name = "Bill Gates" slug = "bill-gates" model_pipeline = "ArpabetGlowTtsMelgan" -melgan = "melgan_vocodes1.0_multispeaker_2020.06.19_12d5838_experiment-2_chkpt-10400.torchjit" glow_tts = "glow-tts_solo_billgates_ljstx_2020.06.24_expr-1_chkpt-2011.torchjit" +melgan = "manyvoice1.0_2020-06-28_12d5838_7050.torchjit" [[speakers]] name = "Mark Zuckerberg" slug = "mark-zuckerberg" model_pipeline = "ArpabetGlowTtsMelgan" -melgan = "melgan_vocodes1.0_multispeaker_2020.06.19_12d5838_experiment-2_chkpt-10400.torchjit" glow_tts = "glow-tts_solo_zuckerberg_ljstx_2020.06.24_expr-1_chkpt-2280.torchjit" +melgan = "manyvoice1.0_2020-06-28_12d5838_7050.torchjit" sample_rate_hz = 21000 # This is lower than normal [[speakers]] name = "Arnold Schwarzenegger (Multi-speaker model)" slug = "arnold-schwarzenegger" model_pipeline = "ArpabetGlowTtsMultiSpeakerMelgan" -melgan = "melgan_vocodes1.0_multispeaker_2020.06.19_12d5838_experiment-2_chkpt-10400.torchjit" glow_tts_multi_speaker = "glow_tts_multispeaker_8-speaker_256-gin_ljs_txlearn_2020.06.19_experiment-5_chkpt-7466.torchjit" speaker_id = 3 +melgan = "melgan_vocodes1.0_multispeaker_2020.06.19_12d5838_experiment-2_chkpt-10400.torchjit" +#melgan = "manyvoice1.0_2020-06-28_12d5838_7050.torchjit" sample_rate_hz = 21000 # This is lower than normal # --------------- Model configs ---------------